Category: technology

Transcript of Lex Fridman interview with Manolis Kellis: Human Genome and Evolutionary Dynamics July 31, 2020

{This is my version of a transcript of the first Lex Fridman interview with Manolis Kellis. I transcribed this conversation because I wanted to dive deeper on a number of topics in this discussion. Perhaps this transcript will be useful to others as well.}

{LF} The following is a conversation with Manolis Kellis, a professor at MIT and head of the MIT Computational biology group. He is interested in understanding the human genome from a computational evolutionary, biological and other cross disciplinary perspectives. He has more big impactful papers and awards that I can list. But most importantly he’s a kind, curious, brilliant human being and just someone I really enjoyed talking to. His passion for science and life in general is contagious. The hours honestly flew by and I’m sure we’ll talk again on this podcast soon. 

{LF [3:50] } …. And now here’s my conversation with Manolis Kellis. What to you is the most beautiful aspect of the human genome?

{MK}    Don’t get me started. So…

{LF} We’ve got time. 

{MK}  The first answer is that the beauty of genomes transcends humanity. So it’s not just about the human genome, genomes in general are amazingly beautiful. And again, I’m obviously biased. So, in my view, the way that I’d like to introduce the human genome and the way that I like to introduce genomics to my classes by telling them, you know, we’re not the inventors of the first digital computer, we are the descendants of the first digital computer. 

Basically life is digital and that’s absolutely beautiful about life. The fact that at every replication step you don’t lose any information because that information is digital. If it was analog, it was just {protein}  concentrations, you’d lose it after a few generations, it would just dissolve away. And that’s what the ancients didn’t understand about inheritance. 

The first person to understand digital inheritance was Mendel of course, and his theory in fact stayed in a bookshelf for like 50 years while Darwin was getting famous about natural selection, but the missing component was this digital inheritance. The mechanism of evolution that Mendel had discovered. So that aspect, in my view, is the most beautiful aspect, but it transcends all of life.

{LF}   And can you elaborate maybe the inheritance part? What was the key thing that the ancients didn’t understand?

{MK}    So the very theory of inheritance as discrete units, you know, throughout the life of Mendel and well after his writing, people thought that his pea experiments were just a little fluke.  That they were just a little exception that would normally not even apply to humans that basically what they saw is: this continuum of eye color, this continuum of skin color, this continuum of hair color, this continuum of height and all of these continuums did not fit with a discrete type of inheritance that Mendel was describing. 

But what’s unique about genomics and what’s unique about the genome is really that there are two copies and that you get a combination of these but for every trait there are dozens of contributing variables. And it was only Ronald Fisher in the 20th century that basically recognized that even five Mendelian Traits would add up to a continuum-like inheritance pattern. And he wrote a series of papers that still are very relevant today about this Mendelian inheritance of continuum-like traits. And I think that that was the missing step in inheritance. So, well before the discovery of the structure of DNA, which is again another amazingly beautiful aspect, the double helix, what I like to call the most noble molecule over time is you know…, holds within it the secret of that discrete inheritance. But the conceptualization of discrete, you know, elements is something that precedes that.

{LF}   So even though it’s discrete, when it materializes itself into actual traits that we see, it can be continuous and basically arbitrarily rich and complex.

{MK}    So if you have five genes that contribute to human height and there aren’t five, there’s 1000. If there’s only five genes and you inherit some combination of them and everyone makes you two inches taller or two inches shorter. It’ll look like a continuous trait, but instead of five, there are thousands and every one of them contributes to less than one millimeter, we change in height more during the day than each of these genetic variants contributes. So by the evening you’re shorter than you were when you woke up.

{LF}   Isn’t it weird then that we’re not more different than we are? Why are we also similar if there’s so much possibility to be different? 

{MK}   Yeah, so there are selective advantages to being medium. If you’re extremely tall or extremely short, you run into selective disadvantages. So you have trouble breathing, you have trouble running, you have trouble sitting, if you’re too tall. If you’re too short, you might, I don’t know, have other selective pressures acting against that. If you look at the natural history of human population, there’s actually selection for height in Northern Europe and selection against height in Southern Europe. So there might actually be advantages to actually being not not super-tall. And if you look across the entire human population, you know, for many many trades, there’s a lot of push towards the middle. Balancing selection is, you know, the usual term for selection that sort of seeks to not be extreme and to sort of have a combination of alleles that sort of, you know, keep recombining. 

And if you look at mate selection super, super tall people will not tend to sort of marry super, super tall people. Very often you see these couples that are kind of compensating for each other and the best predictor of the kids age {height} is very often just take the average of the two parents and then adjust for sex and whom you get it. It’s extremely heritable.

{LF}   Let me ask, you kind of took a step back to the genome outside of just humans. But is there something that you find beautiful about the human genome specifically?

{MK}    So,  I think the genome, if more people understood the beauty of the human genome, there would be so many fewer wars, so much less anger in the world. I mean what’s really beautiful about the human genome is really the variation that teaches us both about individuality and about similarity. So any two people on the planet are 99.9% identical. How can you fight with someone who is 99.9% identical to you? It’s just counterintuitive and yet any two siblings of the same parents differ in millions of locations. So every one of them is basically two to the million {2^1,000,000} unique from any pair of parents, let alone any two random parents on the planet. So that’s I think something that teaches us about the nature of humanity in many ways, that every one of us is as unique as any star and way more unique in actually many ways and yet we’re all brothers and sisters. 

{LF}   And just like stars, most of it is just a fusion reaction.

{MK}    You only have a few parameters to describe stars, you know, 

{LF} Yeah. Exactly.

{MK}  …. mass size, initial size and stage of life. Whereas for humans it’s, you know thousands of parameters scattered across your genome. 

So the other thing that makes humans unique. The other thing that makes inheritance unique in humans is that most species inherit things vertically; basically instinct is a huge part of their behavior. The way that you know….  I mean with my kids we’ve been watching these nest of birds with two little eggs, you know, outside our window for the last few months for the last few weeks as they’ve been growing. And there’s so much behavior that’s hard-coded, birds don’t just learn as they grow. They don’t, you know … there’s no culture like a bird that’s born in Boston will be the same as a bird that’s born in California. So there’s not as much inheritance of ideas, of customs. A lot of it is hard coded in their genome. What’s really beautiful about the human genome is that if you take a person from today and you place them back in ancient Egypt or if you take a person from ancient Egypt and you place them here today, they will grow up to be completely normal. 

That is not genetics. This is the other type of inheritance in humans. So on one hand we have the genetic inheritance which is vertical from your parents down on the other hand we have horizontal inheritance which is the ideas that are built up at every generation are horizontally transmitted. And the huge amount of time that we spend in educating ourselves, a concept known as neoteny, “neo” for newborn and then “teny” for holding. So if you look at humans, I mean the little birds that were eggs two weeks ago and now one of them has already flown off, the other one is ready to fly off.  In two weeks they’re ready to just fend for themselves. Humans:  16 years, 18 years, 24, getting out of college….

{LF}    I’m still learning. So, that’s so fascinating that this picture of a vertical and the horizontal when you talk about the horizontal is in the realm of ideas.

{MK}    Exactly.

{LF}   Okay, so it’s the actual social interactions and ….

{MK}   That’s exactly right. That’s exactly right. So basically the concept of neoteny is that you spend acquiring characteristics from your environment in an extremely malleable state of your brain and the wiring of your brain for a long period of your life. Compared to primates we are useless. You take any primate at seven weeks and any human in seven weeks we lose the battle but at 18 years all bets are off. Like basically our brain continues to develop in an extremely malleable form until very late. And this is what allows education. This is what allows the person from Egypt to do extremely well now. 

And the reason for that is that the wiring of our brain and the development of that wiring is actually delayed. So you know the longer you delay that the more opportunity you have to pass on knowledge, to pass on concepts, ideals, ideas from the parents to the child and what is really absolutely beautiful about humans today is that lateral transfer of ideas and culture. He’s not just from uncles and aunts and teachers at school, but it’s from Wikipedia and review articles on the web and thousands of journals that are sort of putting out information for free and podcasts and videocasts and all of that stuff where you can basically learn about any topic, pretty much everything that would be in any super-advanced textbook in a matter of days instead of having to go to the library of Alexandria and sail there to read three books and then sail for another few days to get to Athens and etc. etc. So the democratization of knowledge and the spread,  the speed of spread of knowledge is what defines I think the human inheritance pattern.

{LF}   So you sound excited about it. Are you also a little bit afraid? Are you more excited by the power of this kind of distributed spread of information? So you put it very kindly that most people are kind of using the internet and you know, looking at Wikipedia,  reading articles, reading papers and so on. But if we are honest, most people online, especially when they’re younger, are probably looking at five second clips on TikTok or whatever the new social network is. Are you,  given this power of horizontal inheritance, are you optimistic or a little bit pessimistic about this new effect of the internet and democratization of knowledge on our …. what would you call this this. gene? Like, would you, would you use the term genome by the way for this?

{MK}    Yeah, I think, you know, we use the genome to talk about DNA, but very often we say, you know, I mean, I’m Greek, so people ask me, hey, what’s in the Greek genome? And I’m like, well, yeah, what’s in the Greek genome is both our genes and also our ideas and our ideals and our culture. 

{LF} The poetic meaning of the word.

{MK}  Exactly. Exactly. Yeah. So I think that there’s a beauty to the democratization of knowledge, the fact that you can reach as many people as you know, any other person on the planet and it’s not who you are, it’s really your ideas that matter, is a beautiful aspect of the internet. 

I think there’s of course a danger of: my ignorance is as important as your expertise. The fact that with this democratization comes the abolishment of respecting expertise, Just because you’ve spent, you know, 10,000 hours of your life studying, I don’t know, human brain circuitry; why should I trust you? I’m just going to make up my own theories and they will be just as good as yours is an attitude that sort of counteract the beauty of the democratization.

And I think that within our educational system and within the upbringing of our children, we have to not only teach them knowledge, but we have to teach them the means to get to knowledge and that, you know, it’s very similar to sort of: You catch a fish for a man for one day, you fed them for one day, you teach them how to fish, you feed him for the rest of their life. 

So instead of just gathering the knowledge they need for any one task, we can just tell them all right: here’s how you Google it, here’s how I figure out what’s real and what’s not, here’s how you check the sources, here’s how you form a basic opinion for yourself. And I think that inquisitive nature is paramount to being able to sort through this huge wealth of knowledge. So you need a basic educational foundation based on which you can then add on the domain specific knowledge, but that basic educational foundation should just, not just the knowledge, but it should also be epistemology, the way to acquire knowledge.

{LF}   I’m not sure any of us know how to do that in this modern day. We’re actually learning. One of the big surprising thing to me about the coronavirus, for example, is that Twitter has been one of the best sources of information, basically like building your own network of experts of of you know, as opposed to the traditional centralized expertise of the WHO {World Health Organization}  and the CDC and the or maybe any one particular respectable person at the top of the department of some kind of institution. You instead look at a you know ten, twenty, hundreds of people; some of whom are young kids with just that are incredibly good at aggregating data and plotting and visualizing that data. That’s been really surprising to me. I don’t know what to make of it. I don’t know how that matures into something stable. You know, I don’t know if you have ideas like what if you were to try to explain to your kids,  where should you go to learn about coronavirus? What would you say?

{MK}    It’s such a beautiful example and I think the current pandemic and the speed at which the scientific community has moved in the current pandemic I think exemplifies this horizontal transfer and the speed of horizontal transfer of information. The fact that, you know, the genome was first sequenced in early January. The first sample was obtained December 29 2019 a week after the publication of the first genome sequence Moderna had already finalized this vaccine design and was moving to production. I mean this is phenomenal,  the fact that we go from not knowing what the heck is killing people in Wuhan to wow, it’s SARS-CoV-2 and here’s the set of genes, here’s the genome, here’s the sequence here, the polymorphisms etc. in the matter of weeks is phenomenal. 

In that incredible pace of transfer of knowledge there have been many mistakes. So you know, some of those mistakes may have been politically motivated or other mistakes may have just been innocuous errors. Others may have been misleading the public for the greater good such as: Don’t wear masks because we don’t want the masks to run out. I mean that was very silly in my view and a very big mistake but the spread of knowledge from the scientific community was phenomenal and some people will point out bogus articles that snuck in and made the front page. Yeah they did. But within 24 hours they were debunked and went out of the front page and I think that’s the beauty of science today. The fact that it’s not, oh, knowledge is fixed. It’s the ability to embrace that nothing is permanent when it comes to knowledge that everything is the current best hypothesis and the current best model that best fits the current data and the willingness to be wrong. The expectation that we’re going to be wrong and the celebration of success based on how long was I not proven wrong for rather than wow, I was exactly right. 

Because no one is going to be exactly right with partial knowledge but the arc towards perfection. I think it’s so much more important then how far you are in your first step and I think that’s what sort of the current pandemic has taught us the fact that you know of course we’re gonna make mistakes but at least we’re going to learn from those mistakes and become better and learn better and spread information better. So if I were to answer the question of where would you go to learn about coronavirus: first textbook? It all starts with the textbook, just opened up a chapter on virology and how coronaviruses work, then some basic epidemiology and sort of how pandemics have worked in the past. What are the basic principles surrounding these first wave? Second wave? Why do they even exist? Then, understanding about growth.

Understanding about the R-naught {R0} and Rt at various time points and then understanding the means of spread: how it spreads from person to person then how does it get into your cells? From when it gets into the cells what are the paths that it takes? What are the cell types that express the particular ACE2 receptor? How is your immune system interacting with the virus? And once your immune system launches its defense, how is that helping or actually hurting your health? What about the cytokine storm? What are most people dying from? Why are the comorbidities and these risk factors even applying? What makes obese people respond more or elderly people respond more to the virus while kids are completely, you know, very often not even aware that they’re spreading it. So, you know, I think there’s some basic questions that you would start from and then I’m sorry to say, but Wikipedia is pretty awesome, Google is pretty awesome. 

{LF}   There used to be a time, maybe five years ago, I forget when but people kind of made fun of Wikipedia for being an unreliable source. I never quite understood. I thought from the early days it was pretty reliable. They’re better than a lot of the alternatives. But at this point it’s kind of like a solid accessible survey paper on every subject ever.

{MK}    There’s an ascertainment bias and a writing bias. So I think this, this is related to people saying so many Nature {journal} papers are wrong and they’re like why would you publish in Nature {journal}? So many Nature {journal} papers are wrong? And my answer is no, no, no. So many Nature {journal} papers are scrutinized and just because more of them are being proven wrong than in other articles is actually evidence that they’re actually better papers overall because they’re being scrutinized at a rate much higher than any other journal. So if you basically judge Wikipedia by not the initial content but by the number of revisions.

{LF} Yeah.

{MK}  Then of course it’s going to be the best source of knowledge. Eventually it’s still very superficial, you then have to go into the review papers etc. But I mean for most scientific project topics, it’s extremely superficial, but it is quite authoritative because it is the place that everybody likes to criticize as being wrong.

{LF}   You say that’s superficial and a lot of topics that I’ve studied a lot of, I find it…  I don’t know if superficial is the right word because superficial kind of implies that is not correct.

{MK}    No, no, I don’t mean any implication of it not being correct, it’s just superficial. It’s basically only scratching the surface. For depth, you don’t go to Wikipedia; you go to the review articles. 

{LF}   But it can be profound in the way that articles rarely …. one of the frustrating things to me about certain computer science, like in the machine learning world articles, they don’t as often take the bigger picture view. You know, there’s a kind of dataset and you show that it works and you kind of show that here’s an architectural thing that creates an improvement and so on and so forth. But you don’t say: well like what does this mean for the nature of intelligence for future datasets we haven’t even thought about or if you’re trying to implement this. Like if we took this data set of 100,000 examples and scale it to 100 billion examples with this method, like look at the bigger picture, which is what a Wikipedia article would actually try to do, which is like: what does this mean in the context of the broad field of computer vision or something like that?

{MK}    Yeah, I agree with you completely but it depends on the topic. I mean for some topics there’s been a huge amount of work, for other topics it’s just a stub. So, you know, I got it. Yeah.

{LF}   Well, yeah, actually, which we’ll talk {about} on genomics was not great.

{MK}    It’s shallow. It’s not wrong. It’s just shallow, shallow. Yeah. Every time I criticize something I should feel partly responsible. Basically if more people from my community went there and edited, it would not be shallow. It’s just that there’s different modes of communication in different fields. And in some fields, the experts have embraced Wikipedia. In other fields, it’s relegated. And perhaps the reason is that if it was any better to start with, people would invest more time. But if it’s not great to start with, then you need a few initial pioneers who will basically go in and say: enough, we’re just going to fix that. and then I think they’ll catch on much more.

{LF}   So if it’s okay before we go on to genomics, can we linger a little bit longer on the beauty of the human genome. You’ve given me a few notes. What else, what else do you find beautiful about the human genome?

{MK}    So the last aspect of what makes the human genome unique in addition to the, you know, similarity and the differences and individuality is that… So very early on people would basically say, oh you don’t do that experiment in humans. You have to learn about that in fly or you have to learn about that in yeast first or in mouse first or in a primate first. {These are examples of model organisms} And the human genome was in fact relegated to sort of the last place that you’re going to go to learn something new that has dramatically changed. And the reason that changed is human genetics. We are the species on the planet that’s the most studied right now. It’s embarrassing to say that, but this was not the case a few years ago. Used to be, you know, first viruses, then bacteria, then yeast, then the fruit fly, then the worm, then the mouse and eventually human was very far last.

{LF}   So it’s embarrassing that it took us this long to focus on it or they ….

{MK}    It’s embarrassing that the model organisms have been taken over because of the power of human genetics. That right now it’s actually simpler to figure out the phenotype of something by mining this massive amount of human data then by going back to any of the other species. And the reason for that is that if you look at the natural variation that happens in a population of seven billion; you basically have a mutation in almost every nucleotide. So every nuclear that you want to perturb; you can go find a living breathing human being and go test the function that nucleotide by sort of searching the database and finding that person. 

{LF}   Wait, why is that embarrassing? It’s a beautiful dataset.

{MK}    It’s embarrassing for the model organism. 

{LF}   For the flies?

{MK}  Yeah, exactly. 

{LF} I mean, do you feel, on a small tangent, is there something of value in the genome of a fly and other these model organisms that you miss,  that we wish we would have would be looking at deeper?

{MK}    So directed perturbation of course. So I think the place where humans are still lagging is the fact that in an animal model you can go and say, well, let me knockout this gene completely.

{LF} Got it.

{MK}  … and let me knockout these three genes completely. And at the moment you get into combinatorics, it’s something you can’t do in humans because they’re just simply aren’t enough humans on the planet.  

And again, let me be honest, we haven’t sequenced all seven billion people. It’s not like we have every mutation, but we know that there’s a carrier out there. 

[29:15] So if you look at the trend and the speed with which human genetics has progressed, we can now find thousands of genes involved in human cognition, in human psychology, in the emotions and the feelings that we used to think are uniquely learned. It turns out there’s a genetic basis to a lot of that. 

So the … you know, the human genome has continued to elucidate through these studies of genetic variation so many different processes that we previously thought were, you know, something that like free will.  Free will is this beautiful concept that humans have had for a long time? You know, in the end, it’s just a bunch of chemical reactions happening in your brain and the particular abundance of receptors that you have this day based on what you ate yesterday or that you have been wired with based on, you know, your parents and your upbringing etc. determines a lot of that quote unquote free will components to, you know, sort of narrower and narrower, you know, sort of slices.

{LF}   So on that point, how much freedom do you think we have to escape the constraints of our genome? 

{MK}  {laughs}

{LF}  You’re making it sound like more and more, we’re discovering that our genome is actually…. has a lot of the story already encoded into it. How much freedom do we have, do you think?

{MK}    So let me describe what that freedom would look like. That freedom would be my saying: Oh, I’m going to resist the urge to eat that apple because I choose not to. But there are chemical receptors that made me not resist the urge to prove my individuality and my free will by resisting the apple. So then the next question is well, maybe now I’ll resist the urge to resist the apple and I’ll go for the chocolate instead to prove my individuality. But then what about those other receptors that you know….

{LF}   That might be all encoded in there.

{MK}    So it’s kicking the bucket down the road and basically saying, well your choice will …. may have actually been driven by other things that you actually are not choosing. So that’s why it’s very hard to answer that question. 

{LF}   It’s hard to know what to do with that. I mean, if if the genome has …. if if there’s not much freedom, it’s

{MK}    It’s the butterfly effect. It’s basically that in the short term you can predict something extremely well by knowing the current state of the system. But a few steps down, it’s very hard to predict based on the current knowledge. Is that because the system is truly free?

 When I look at weather patterns that can predict the next 10 days. Is it because the weather has a lot of freedom and after 10 days it chooses to do something else? Or is it because in fact the system is fully deterministic and there’s just a slightly different magnetic field of the Earth, slightly more energy arriving from the sun, a slightly different spin of the gravitational pull of Jupiter that is now causing all kinds of tides and slight deviation of the moon etc. Maybe all of that can be fully modeled. Maybe the fact that China is emitting a little more carbon today is actually going to affect the weather in Egypt in three weeks and all of that could be fully modeled.

 In the same way. If you take a complete view of a human being now, you know, I model everything about you. The question is, can I predict your next step probably, but how far and if it’s a little further is that because of stochasticity and sort of chaos properties of unpredictability of beyond a certain level or was that actually true free will?

{LF}   Yeah. Yeah. So the number of variables might be so…. you might need to build an entire universe.

{MK}    To simulate a human and then maybe that human will be fully simulatable but maybe aspects of free will will exist. And where is that free will coming from? It’s still coming from the same neurons or maybe from a spirit inhabiting these neurons. But again, you know, it’s very difficult empirically to sort of evaluate where does free will begin and sort of chemical reactions and electric signals end.

{LF}   So on that topic, let me ask the most absurd questions that most MIT faculty roll their eyes on.  {laugh} But what do you think about the simulation hypothesis and the idea that we live in a simulation?

{MK}    I think it’s complete BS.

{LF}  Okay  {laugh}

{MK}   {laugh} There’s no empirical evidence; absolutely none.

{LF}   Not in terms of empirical evidence or not. But in terms of thought experiment, does it help you think about the universe? I mean, if you look at the genome, it’s encoding a lot of the information that is required to create some of the beautiful human complexity that we see around us. It’s an interesting thought experiment. How much, you know, parameters do we need to have in order to model some, you know, this full human experience? Like if you wanted to build a video game, how hard it would be to build a video game that’s convincing enough and fun enough and you know it has a consistent laws of physics, all that stuff. It’s not interesting to you as a thought experiment?.  {laugh} 

{MK}    I mean it’s cute but you know, it’s Occam’s razor. I mean what’s more realistic: the fact that you’re actually a machine or that you’re you know, a person, what’s you know the fact that all of my experiences exist inside the chemical molecules that I have or that somebody is actually, you know, simulating all that.

{LF}   Well you did refer to humans as a digital computer earlier.

{MK}    Of course, of course, but that does not 

{LF} It’s kind of a machine. Right? 

{MK}  I know. I know but I think the probability of all that is nil and let the machines wake me up and just terminate me now if it’s not, I challenge you machines.

{LF}   They’re going to wait a little bit to see what you’re going to do next. It’s fun. It’s fun to watch, especially the clever humans. What’s the difference to you between the way a computer stores information and the human genome stores information. So you also have roots and your work …. would you say you’re when you introduce yourself at a bar,

{MK}   It depends who I’m talking to. 

{LF}    {laugh}  Would you say it’s computational biology? Do you reveal your expertise in computers?

{MK}    It depends who I’m talking to truly. I mean basically if I meet someone who’s in computers, I’ll say, oh, I mean professor of computer science, if I meet someone who’s in engineering as a computer science and electrical engineering, if I meet someone in biology and say, hey, I work on genomics. If I meet someone in medicine, like hey, I work on, you know, genetics.

{LF}   You’re a fun person to meet at a bar, I got you. But so

{MK}   No, no, but what I’m trying to say is that I don’t…. I mean there’s no single attribute that will define myself as, you know, there’s a few things I know there’s a few things I studied. There’s a few things I have degrees on and there’s a few things that I grant degrees in. And you know, I, I publish papers across the whole gamut, you know, the whole spectrum of computation to biology etc. I mean,  the complete answer is that I use computer science to understand biology. So you know, I develop methods in AI, machine learning,  statistics in algorithms etc. But the ultimate goal of my career is to really understand biology. If these things don’t advance our understanding of biology. I’m not as fascinated by them. Although there are some beautiful computational problems by themselves. I’ve sort of made it my mission to apply the power of computer science to truly understand the human genome, health, disease, you know and the whole gamut of how our brain works, how our body works and all of that. Which is so fascinating. 

{LF}   So the dream, there’s not an equivalent sort of complimentary dream of understanding human biology in order to create an artificial life and artificial brain and an artificial intelligence that supersedes the intelligence and the capabilities of us humans.

{MK}    It’s an interesting question. It’s a fascinating question. So understanding the human brain is undoubtedly coupled to how do we make better AI because so much of AI has in fact been inspired by the brain. It may have taken fifty years since the early days of neural networks until we have you know, all of these amazing progress that we’ve seen with you know deep belief networks and you know all of these advances in Go and chess, in image synthesis, in deep fakes, in you name it. But the underlying architecture is very much inspired by the human brain which actually posits a very, very interesting question. Why are neural networks performing so well? And they perform amazingly well. Is it because they can simulate any possible function? And the answer is no, no. They simulate a very small number of functions. Is it because they can simulate every possible function in the universe? And that’s where it gets interesting. The answer is actually a little closer to that, and here’s where it gets really fun.

If you look at the human brain and human cognition, it didn’t evolve in a vacuum. It evolved in a world with physical constraints like the world that inhabits this, it is the world that we inhabit. And if you look at our senses, what do they perceive? They perceive different parts of the electromagnetic spectrum? You know? The hearing is just different movements in air, the touch, etc. I mean all of these things, we’ve built the intuitions for the physical world that we inhabit and our brains and the brains of all animals evolved for that world. 

And the AI systems that we have built happen to work well with images of the type that we encounter in the physical world that we inhabit. Whereas if you just take noise and you add a random signal that doesn’t match anything in our world, neural networks will not do as well. And that actually basically has this whole loop around this, which is this was designed by studying our own brain which was evolved for our own world and they happened to do well in our own world and they happen to make the same types of mistakes that humans make many times. And of course you can engineer images by adding just the right amount of, you know, sort of pixel deviations to make a zebra look like a baboon and stuff like that or like a table. But ultimately the undoctored images, at least, are very often mistaken. I don’t know between muffins and dogs, for example, in the same way that humans make those mistakes.

So it’s you know, there’s no doubt in my view that the more we understand about the tricks that our human brain has evolved to understand the physical world around us, the more we will be able to bring new computational primitives in our AI systems to again better understand, not just the world around us, but maybe even the world inside us and maybe even the computational problems that arise from new types of data that we haven’t been exposed to but are yet inhabiting the same universe that we live in with the very tiny little subset of functions from all possible mathematical functions.

{LF}   Yeah. And that small subset of function is all that matters to us, humans really, that’s what makes….

{MK}   It’s all that has mattered so far. And even within our scientific realm, it’s all that seems to continue to matter. But I mean, I always like to think about our senses and how much of the physical world around us we perceive. And if you look at the LIGO experiment over the last, you know, year and a half has been all over the news, what did LIGO do? It created a new sense for human beings, a sense that has never been sensed in the history of our planet. Gravitational waves have been traversing the earth since its creation a few billion years ago. 

Life has evolved senses to sense things that were never before sensed light was not perceived by early life, no one cared. And eventually photoreceptors evolved and the ability to sense colors by catching different parts of that electromagnetic spectrum and hearing evolved and touch evolved etc. But no organism evolved a way to sense neutrinos floating through earth, or gravitational waves flowing through earth etc. And I find it so beautiful in the history of not just humanity but life on the planet, that we are now able to capture additional signals from the physical world than we ever knew before. And { unintelligible } for example have been all over the news in the last few weeks. The concept that we can capture and perceive more of that physical world is as exciting as the fact that we are, we were blind to it is traumatizing before. Because that also tells us, you know, we’re in 2020 Picture yourself in 3020. Or in 20…

{LF}   What new senses, what might we discover?

{MK}    Is it, you know, could it be that we’re missing 9/10 of physics?

{LF}   Most of physics…

{MK}    That like there’s a lot of physics out there that we’re just blind to. Completely oblivious to it. And yet they’re permeating us all the time. 

{LF}   Yes, it might be right in front of us. 

{MK}    So, when you’re thinking about premonitions,

{LF}   Yeah.

{MK}    A lot of that is ascertainment bias {Sampling bias}. Like yeah, you know, every now and then you’re like, oh I remember my friend and then my friend doesn’t appear and I’ll forget that I remember my friend, but every now and then my friend will actually appear and like, oh my God, I thought about you a minute ago, you just called me. That’s amazing. So, you know, some of that is this, but some of that might be that there are within our brain sensors for waves that we admit that we’re not even aware of. And this whole concept of when I hug my children, there’s such an emotional transfer there that we don’t comprehend. I mean sure, yeah, of course, we’re all hardwired for all kinds of touchy feely things between parents and kids, it’s beautiful. Between partners is beautiful and tender. But then there are intangible aspects of human communication that I don’t think it’s unfathomable that our brain has actually evolved ways and sensors for it that we just don’t capture. We don’t understand the function of the vast majority of our neurons and maybe our brain is already sensing it, but even worse, maybe our brain is not sensing it at all and were oblivious to this until we build a machine that suddenly is able to sort of capture so much more of what’s happening in the natural world. 

{LF}   So what you’re saying is we … physics is going to discover a sensor for love. {laughs}

{MK}    And maybe dogs are off scale for that. {laughs} And we’ve been, you know, we’ve been oblivious to it the whole time because we didn’t have the right sensor and now you’re going to have a little wrist {sensor}  that says, oh my God, I feel all this love in the house. I sense some disturbance in the force.

{LF}   And cats will have zero. None. {laughs}

{MK}    {laughs} No, none.

{LF}   It’s just me. But let’s take a step back to unfortunately ….

{MK}    To one of the four hundred topics that we had actually planned. 

{LF}   But to our sad time in 2020 when we only have just a few sensors and very primitive early computers. So you have a foot in computer science and a foot in biology. In your sense: How do computers represent information differently than like the genome or biological systems?

{MK}    So, first of all, let me correct that no, we’re in an amazing time in 2020. Computer science is totally awesome. And physics is totally awesome. And we have understood so much of the natural world than ever before. So I am extremely grateful and feeling extremely lucky to be living in the time that we are because, you know, first of all, who knows when the asteroid will hit? And second, you know, of all times in humanity, this is probably the best time to be a human being. And this might actually be the best place to be a human being. So anyway, you know, for anyone who loves science, this is it, this is awesome. It’s a great time

{LF}   At the same time, just a swift comment. All I meant is that if you look several hundred years from now and we end up somehow not destroying ourselves.

{MK}  Yeah. 

{LF} People probably look back at this time in computer science and at your work of Manolis at MIT

{MK}   As infantile….

{LF}  As infantile and silly and how ignorant it all was.

{MK}    I like to joke very often with my students that, you know, we’ve written so many papers, were published so much, we’ve been cited so much and every single time I tell my students you know, the best is ahead of us. What we’re working on now is the most exciting thing I’ve ever worked on. So in a way, I do have this sense of, yeah, even the papers I wrote ten years ago, they were awesome at the time. But I’m so much more excited about where we’re heading now. And I don’t mean to minimize any of the stuff we’ve done in the past, but you know, there’s just this sense of excitement about what you’re working on now, that as soon as the paper is submitted, it’s like, it’s old. You know, I can’t talk about that anymore. 

{LF}   At the same time, you’re not… you probably are not going to be able to predict what are the most impactful papers and ideas. When people look back two hundred years from now at your work, what would be the most exciting papers? And it may very well be not the thing that you expected or…. 

{MK}  Yeah.

{LF} …. or the things you’ve got awards for or …..

{MK}    You know, that between some fields, I don’t know, I feel slightly differently about it in our field. I feel that, I kind of know what are the important ones and there’s a very big difference between what the press picks up on and what’s actually fundamentally important for the field. And I think for the fundamentally important ones, we kind of have a pretty good idea what they are. And it’s hard to sometimes get the press excited about the fundamental advances, but, you know, we take what we get and celebrate what we get. And sometimes, you know, one of our papers which was in a minor journal, made the front page of Reddit and suddenly had like hundreds of thousands of views, even though it was in a minor journal, because, you know, somebody pitched it the right way that it suddenly caught everybody’s attention. Whereas other papers that are sort of truly fundamental. You know, we have a hard time getting the editors even excited about them when so many hundreds of people are already using the results and building upon them. So I do appreciate that there’s a discrepancy between the perception and the perceived success and the awards that you get for various papers. But I think that fundamentally, you know, that, you know, some paper…. So, so when you’re right,

{LF}   Is there a paper you’re most proud of, you know, now you just trapped yourself?

{MK}    No, no, no, no. I mean

{LF}   Is there a line of work that you have a sense is really powerful that you’ve done to date? You’ve done so much work in so many directions, which is interesting. Is there something where you think it is quite special?

{MK}    I mean it’s like asking me to say which of my three children I love best. I mean…. 

{LF} Exactly. 

{MK}  So, I mean, it’s such a gimme question that it is so difficult not to brag about the awesome work that my team and my students have done. And I’ll just mention a few off the top of my head. I mean basically there’s a few landmark papers that I think have shaped my scientific path. And you know, I like to somehow describe it as a linear continuation of one thing led to another, led to another, led to another. And you know, it kind of all started with skip skip, skip, skip skip. Let me try to start somewhere in the middle. 

So my first PhD paper was the first comparative analysis of multiple species. So, multiple complete genomes. {Note 1} So for the first time we basically developed the concept of genome wide evolutionary signatures. The fact that you could look across the entire genome and understand how things evolve. And from these signatures of evolution, you could go back and study any one region and say that’s a protein coding gene, that’s a RNA gene, that’s a regulatory motif, that’s a binding site. And so on and so forth. 

{LF}   Sorry, so comparing different ….

{MK}    Different species.

{LF}  Species of the same….

{MK}  Take human, mouse, rat and dog, you know, they’re all animals, they’re all mammals. They’re all performing similar functions with their heart, with their brain, with their lungs, etc, etc. So there’s many functional elements that make us uniquely mammalian and those mammalian elements are actually conserved. 99% of our genome does not code for protein, 1% codes for protein. The other 99%, we frankly didn’t know what it does until we started doing this comparative genomic studies. So basically these series of papers in my career have basically first developed that concept of evolutionary signatures and then applied them to yeast, applied them to flies, applied them to four mammals, applied them to seventeen fungi, applied them to twelve Drosophila {fruit fly} species, applied them to then 29 mammals and now 200 mammals.

{LF}   So sorry. So can we ….So the evolutionary signatures, it seems like it’s such a fascinating idea. I’m probably going to linger on your early PhD work for two hours? But what is, how can you reveal something interesting about the genome by looking at the multiple, multiple species and looking at the evolutionary signatures?

{MK}    Yeah. So, you basically align the matching regions. So everything evolved from a common ancestor way, way back and mammals evolved from a common ancestor about 60 million years back. So after you know, the meteor that killed off the dinosaurs landed ….

{LF} Allegedly.

{MK}  …. near Machu Picchu, we know the {Chicxulub} crater. It didn’t allegedly land.

{LF} That was the aliens. 

{MK}  Okay, just slightly north of Machu Picchu {Yucatan} in the Gulf of Mexico. There’s a giant hole that, that meteor,

{LF}   By the way, sorry, is that definitive? Do people, have people, conclusively figured out what killed the dinosaurs?

{MK}   I think so.

{LF}  So it was a meteor?

{MK}    Well, you know, for volcanic activity, all kinds of other stuff is coinciding, but the meteor is pretty unique and we now have…. 

{LF} That’s also terrifying, we still have a lot of 2020 left… 

{MK}    No, but think about it this way. So the dinosaurs ruled the earth for 175 million years. We humans have been around for what? Less than one million years if you’re super generous about what you call humans and you include chimps basically. So, we are just getting warmed up and you know, we’ve ruled the planet much more ruthlessly than Tyrannosaurus rex. {laughs} T. rex had much less of an environmental impact than we did. And if you give us another 174 million years, you know, humans will look very different if we make it that far. So I think dinosaurs basically are much more of life history on earth and we are in all respects. But look at the bright side when they were killed off another life form emerged: mammals.

{LF}   And that’s that whole…  the evolutionary branching that’s happened. So you kind of have when you have these evolutionary signatures they’re basically a map of how the genome changed.

{MK}    Exactly, exactly. So, now you can go back to this early mammal that was hiding in caves and you can basically ask what happened after the dinosaurs were wiped out. A ton of evolutionary niches opened up and the mammals started populating all of these niches. And in that diversification there was room for expansion of new types of functions. So some of them populated the air with bats flying, a new evolution of flight. Some populated the oceans with dolphins and whales going out to swim etc. 

But we all are fundamentally mammals. So you can take the genomes of all these species and align them on top of each other and basically create nucleotide resolution correspondences. What my PhD work showed is that when you do that, when you line up species on top of each other, you can see that within protein coding genes, there’s a particular pattern of evolution that is dictated by the level at which evolutionary selection acts. If I’m coding for a protein and I changed the third codon position of a triplet that codes for the amino acid; the same amino acid will be encoded. So that basically means that any kind of mutation that preserves that translation, that is invariant to that ultimate functional assessment that evolution will give, is tolerated.

 So for any function that you’re trying to achieve there is a set of sequences that encode it. You can now look at the mapping the,  you know, the graph isomorphism if you wish, between all of the possible DNA encodings of a particular function and that function. And instead of having just that exact sequence at the protein level you can think of the set of protein sequences that all fulfill the same function. What’s evolution doing? Evolution has two components. One component is random, blind and stupid mutation. The other component is super smart, ruthless selection. 

{phone rings} Yeah that’s my mom calling from Greece. Yes. I might be a fully grown man. Yes but I am Greek.

{LF}   Did you just cancel the call? 

{MK}    I know I’m in trouble. She’s going to be calling the cops.

{LF}   I’m going to edit this clip out and send it to her.

{laughter}

{LF}   So there’s a lot of encoding for the same kind of functions.

{MK}    So you now have this mapping between all of the set of functions that could all encode the same, all of the set of sequences that can all encode the same function. What evolutionary signatures does is that it basically looks at the shape of that distribution of sequences that all encode the same thing. And based on that shape you can basically say: “Oh proteins have a very different shape than RNA structures, than regulatory motifs etc. So just by scanning a sequence, ignoring the sequence and just looking at the pattern of change, I’m like wow this thing is evolving like a protein and that thing is evolving like a motif and that thing is evolving. So that’s exactly what we just did for COVID. So our paper that we posted in bioRxiv about coronavirus basically took this concept of evolutionary signatures and applied it on the SARS-CoV-2 genome that is responsible for the COVID-19 pandemic.

{LF}   And comparing it to…. 

{MK}    To 44  Sarbecovirus species. 

{LF} – What word did you just use?

{MK}  – Sarbecovirus. The SARS related beta coronavirus, it’s a portmanteau. 

{LF} – So that’s a family {subgenus} of viruses. 

{MK}  – Yeah.

{LF} – How big is that family? 

{MK}  -We have 44 species that are 

{LF} – 44 species? Viruses are clever.

{MK}  – but no no, but there’s just 44 again, we don’t call them species in viruses, we call them strains. But anyway, there’s 44 strains and that’s a tiny little subset of, you know, maybe another 50 strains that are just far too distantly related. Most of those only infect bats as a host and a subset of only four or five have ever infected humans. And we basically took all of those and we align them in the same exact way that we’ve aligned mammals. And then we looked at what proteins are, you know, which of the currently hypothesized genes for the coronavirus genome are in fact evolving like proteins and which ones are not. And what we found is that ORF10,  the last little open reading frame, the last little gene in the genome is bogus. That’s not a protein at all.

{LF} What is it?

{MK}  It’s an RNA structure.

{LF}   That doesn’t have…..

{MK}    It doesn’t get translated into amino acids.

{LF}   And that’s so it’s important to narrow down to basically discover what’s useful and what’s not.

{MK}    Exactly. Basically what are …. what is even the set of genes. The other thing that these evolutionary signatures showed is that within ORF3A, like a tiny little additional gene encoded within the other gene. So you can translate the DNA sequence in three different reading frames. If you start in the first one, it’s ATG cetera. If you start on the second is TGC et cetera. And there’s a gene within a gene. So there’s a whole other protein that we didn’t know about, that might be super important. So we don’t even know the building blocks of SARS-CoV-2. So if we want to understand coronavirus biology and eventually fight it successfully, we need to even have the set of genes and these evolutionary signatures that are developed in my PhD work, we just recently used.

{LF}   You know what, let’s run with that tangent for a little bit, if it’s okay. Can we talk about the COVID-19 a little bit more like how ….. what’s your sense about the genome, the proteins, the functions that we understand about COVID-19. Where do we stand in your sense? What are the big open problems? And also, you know, you kind of said it’s important to understand what are like the important proteins and like why is that important?

{MK}    So what else does the comparison of the species tell us?  What it tells us is, how fast are things evolving? It tells us about at what level is the acceleration or deceleration pedal set, for every one of these proteins. So the genome has 30 genes, some genes evolve super, super fast. Others evolve super, super slow. If you look at the polymerase gene that basically replicates the genome, that’s a super slow evolving one. If you look at the nucleocapsid protein, it’s also super slow evolving. If you look at the Spike1 {Spike subunit S1} protein, this is the part of the spike protein that actually touches the ACE2 receptor as it enables the virus to attach to your cells.

{LF}   That’s the thing that gives it that visual

{MK}    The corona-look basically. So, the spike protein sticks out of the virus. And there’s the first part of the protein, S1, which basically attaches to the ACE2 receptor. And then S2 is the latch that sort of pushes and channels the fusion of the membranes and then the incorporation of the viral RNA inside our cells which then gets translated into all of these 30 proteins. So the S1 protein is evolving ridiculously fast. So if you look at the stop {brake pedal} versus the gas pedal, the gas pedal is all the way down. ORF8 is also evolving super fast and ORF6 is evolving super fast. We have no idea what they do. We have some idea but nowhere near what S1 is. So what the …

{LF}   Isn’t that really terrifying that S1 is evolving? That means that’s a really useful function. If it’s evolving fast, does that mean new strengths can be created? Or does something….

{MK}    THat means that it’s searching for how to match, how to best match the host? So basically anything, in general, in evolution, if you look at genomes, anything that’s contacting the environment is evolving much faster than anything that’s internal. And the reason is that the environment changes. So if you look at the evolution of the sarbecoviruses, the S1 protein has evolved very rapidly because it’s attaching to different hosts each time. We think of them as bats but there’s thousands of species of bats and to go from one species of bat to another species of bat….

{LF} You have to figure out …. 

{MK}  S1 to the new ACE2 receptor that you’re going to be facing in that new species.

{LF}   [1:03:00] Quick tangent. Is it fascinating to you that viruses are doing this? I mean it feels like they’re this intelligent organism. I mean, is it like …. does that give you pause how incredible it is that the evolutionary dynamics that you’re describing is actually happening and they’re freaking out figuring out how to jump from bats to humans all in this distributed fashion. And then most of us don’t even say they’re alive or intelligent or whatever.

{MK}    So intelligence is in the eye of the beholder, you know, stupid is as stupid does, as Forrest Gump would say and intelligent is as intelligent does. So basically if the virus is finding solutions that we think of as intelligent yeah, they’re probably intelligent but that’s again in the eye of the beholder. 

{LF} Do you think viruses are intelligent? 

{MK}  Of course not.

{LF} Really? 

{MK}  No, because …

{LF} It’s so incredible. 

{MK} : Remember when I was talking about the two components of evolution, one is the stupid mutation which is completely blind and the other one is the super smart selection which is ruthless. So it’s not viruses who are smart, it’s this component of evolution that is smart. So it’s evolution that sort of appears smart and how is that happening? By huge parallel search across thousands of you know, parallel infections throughout the world right now. 

{LF}   Yes, but so to push back on that. So yes. So then the intelligence is in the mechanism but then by that argument, viruses would be more intelligent because there’s just more of them. So the search they’re basically …. the brute force search that’s happening with viruses because there’s so many more than than humans then they’re taken as a whole are more intelligent. I mean, so you don’t think it’s possible that, I mean, who runs…. Would we even be here if viruses weren’t? I mean who runs this thing?

{MK}    So the virus… So let me answer,let me answer your question. So we would not be here if it wasn’t for viruses. And part of the reason is that if you look at mammalian evolution early on in this mammalian radiation that basically happened after the death of the dinosaurs is that some of the viruses that we had in our genome spread throughout our genome and created binding sites for new classes of regulatory proteins and these binding sites that landed all over genome are now control elements that basically control our genes and sort of help the complexity of the circuitry of mammalian genomes. So, you know, everything is coevolution and ….

{LF}   We’re working together.

{MK}    Yeah.

{LF} And yet you said they’re dumb. 

{MK}  No, I never said they’re dumb. They just don’t care. They don’t care. Another thing: Oh is the virus trying to kill us? No, it’s not. The virus is not trying to kill you. It’s not, it’s actually actively trying to not kill you. So when you get infected if you die: “bummer I killed him”, is what the reaction of the virus will be. Why? Because that virus won’t spread. Many people have a misconception that viruses are smart or viruses are mean. They don’t care. It’s like you have to clean yourself of any kind of anthropomorphism out there. 

{LF} I don’t know. 

{MK}  Oh yes. 

{LF}   So there’s a sense when taken as a whole that there’s a …..

{MK}    It’s in the eye of the beholder. Stupid is as stupid does. Intelligence is as intelligence does. So, if you want to call them intelligent, that’s fine because the end result is that they’re finding amazing solutions.

{LF} Right. 

{MK}  I mean I am in awe …

{LF} But they’re so dumb about it. They’re just doing dumb….

{MK}  They don’t care. They’re not dumb and they’re not even..

{LF} Sorry. They don’t care.

{MK}  Exactly.

{LF}  The care word is really interesting. 

{MK}  Exactly.

{LF} I mean it could be an argument that they are conscious.

{MK}   They’re just dividing. They’re not {conscious}, they’re just dividing. They’re just a little entity which happens to be dividing and spreading. It just doesn’t want to kill us. In fact it prefers not to kill us. It just wants to spread. And when I say wants, again I’m anthropomorphizing but it’s just that if you have two versions of a virus, one acquires a mutation that spreads more, that’s going to spread more. One acquires a mutation that spreads less,that’s going to be lost. 

{LF} Yeah.

{MK}  One acquires a mutation that enters faster, that’s gonna be kept. One acquires a mutation that kills you right away, it’s going to be lost. So, over evolutionary time, the viruses that spread super well but don’t kill the host are the ones that are going to survive.

{LF}   Yeah but so … you’ve brilliantly described the basic mechanisms of how it all happens. But when you zoom out and you see the, you know,  the entirety of viruses, maybe across different strains of viruses. It seems like a living organism.

{MK}    I am in awe of biology. I find biology amazingly beautiful. I find the design of the current coronavirus, however lethal it is, amazingly beautiful: the way that it is encoded, the way that it tricks your cells Into making thirty proteins from a single RNA. Human cells don’t do that. Human cells make one protein from each RNA molecule. They don’t make two, they make one. We are hardwired to make only one protein from every RNA molecule. 

And yet this virus goes in and throws in a single messenger RNA. Just like any messenger RNA. We have tens of thousands of messenger RNAs in ourselves at any one time in every one of ourselves. It throws in one RNA and that RNA is,so I’m going to use your word here, not my word: “intelligent”; that it hijacks the entire machinery of your human cell. It basically has at the beginning a giant open reading frame. That’s a giant protein that gets translated.  Two-thirds of that RNA makes a single giant protein. That single protein is basically what a human cell would make.  It’s like: “Oh here’s a start codon and I’m going to start translating here.” Human cells are kind of dumb. I’m sorry again this is not the word that we normally use but the human cell is basically: “So, this is an RNA, it must be mine. Let me translate it.” 

And it starts translating it and then you’re in trouble. Why? Because that one protein as it’s growing gets cleaved into about 20 different peptides. The first peptide and the second peptide start interacting and the third one in the fourth one and they shut off the ribosome of the whole cell to not translate human RNAs anymore. So the virus basically hijacks your cells and it cuts, it cleaves every one of your human RNAs to basically say to the rivals: “Don’t translate this one, junk. Don’t look at this one, junk.” And it only spares its own RNA because they have a particular mark that it spares. 

Then all of the ribosomes that normally make protein in your human cells are now only able to translate viral RNA like more and more and more and more of them. That’s the first 20 proteins. In fact, halfway down, about protein 11 – between 11 and 12, you basically have a translational slippage where the ribosomes skip {the} reading frame and it translates from one reading frame to another reading frame. That means that about half of them are gonna be translated from 1-11 and the other half are going to be translated from 12 to 16. It’s gorgeous. And then you’re done. Then that mRNA will never translate the last ten proteins. But spike is the one right after that one. So how does spike even get translated? 

This positive-strand RNA virus has a reverse transcriptase which is an RNA based reverse transcriptase. So from the RNA on the positive strand it makes an RNA on the negative strand and in between every single one of these genes, these open reading frames, there’s a little signal AACGCA, or something like that that basically loops over to the beginning of the RNA. And basically, instead of sort of having a single full negative strand RNA, it basically has a partial negative strand RNA that ends right before the beginning of that gene. And another one that ends right before the beginning of that {uses hands to indicate a subsequent} gene. These negative strand RNAs now make positive strand RNA. That then look to the human whole cell just like any other human mRNA. It’s like: “Oh great I’m going to translate that one,” because it doesn’t have the cleaving that the virus has now put on all your human genes and then you’ve lost the battle. That cell is now only making proteins for the virus that will then create the spike protein, the envelope protein, the membrane protein, the nucleocapsid protein that will package up the RNA and then sort of create new viral envelopes. And these will then be secreted out of that cell in new little packages that will then infect the rest of cells.

{LF} Repeat the whole process. 

{MK}  It’s beautiful, right? It’s mind blowing..

{LF}   It’s hard not to anthropomorphize but …. {both laughing}

{MK}    It’s so gorgeous.

{LF}   So there is a beauty to it. Is it terrifying to you?

{MK}    So this is something that has happened throughout history. Humans have been nearly wiped out over and over and over again and yet never fully wiped out. So I’m not concerned about the human race, I’m not even concerned about, you know,the impact on sort of our survival as a species.This is absolutely something, I mean, you know, human life is so invaluable and every one of us is so invaluable. But if you think of it as sort of, is this the end of our species by no means.

So basically, so let me explain. The Black Death killed what, 30% of Europe? That has left a tremendous imprint, you know, a huge hole, horrendous hole in the genetic makeup of humans. There’s been a series of wiping out of huge fractions of entire species or just entire species altogether and that has a consequence on the human immune repertoire. 

If you look at how Europe was shaped and how Africa was shaped by malaria for example, all the individuals that carry imitation that protected from malaria were able to survive much more. And if you look at the frequency of sickle cell disease and the frequency of malaria, the maps are actually showing the same pattern, the same imprint on Africa and that basically led people to hypothesize that the reason why sickle cell disease is so much more frequent in Americans of African descent is because there was selection in Africa against malaria leading to sickle cell, because when the cells sickle, malaria is not able to replicate inside yourselves as well and therefore you protect against that. 

So if you look at human disease, all of the genetic associations that we do with human disease, you basically see the imprint of these waves of selection, killing off gazillions of humans. And there’s so many immune processes that are coming up as associated with so many different diseases.

The reason for that is similar to what I was describing earlier, where the outward facing proteins evolved much more rapidly because the environment is always changing. But what’s really interesting {in}  the human genome is that we have co-opted many of these immune genes to carry out non-immune functions. For example, in our brain we use immune cells to cleave off neuronal connections that don’t get used. This whole “use it or lose it;” we know the mechanism it’s microglia to cleave off neuronal synaptic connections that are just not utilized. When you utilize them, you mark them in a particular way that basically when the microglia come tell it:  “Don’t kill this one, it’s used now.” and the microglia will go off and kill the ones you don’t use. This is an immune function which is co-opted to do non-immune things. 

If you look at our adipocytes, M1 versus M2 macrophages  inside our fat will basically determine whether you’re obese or not. And these are again immune cells that are resident and living within these tissues. So, many disease associations …..

{LF}   That we co opt these kinds of things for incredibly complicated functions.

{MK}   Exactly. Evolution works in so many different ways which are all beautiful and mysterious.

{LF}  But not intelligent.

{MK}  Not intelligent, it’s in the eye of the beholder. But the point that I’m trying to make is that if you look at the imprint that COVID will have, hopefully it will not be big. Hopefully the US will get its act together and stop the virus from spreading further. But if it doesn’t, it’s having an imprint on individuals who have particular genetic repertoires. So if you look at now the genetic associations of blood type and immune function cells, etc. there’s actually association, genetic variation that basically says how much more likely am I or you to die if we contract the virus. And it’s through these rounds of shaping the human genome that humans have basically made it so far. 

And selection is ruthless and it’s brutal and it only comes with a lot of killing. But this is the way that viruses and environments have shaped the human genome. Basically when you go through periods of famine you select for particular genes. And what’s left is not necessarily better; it’s just whatever survived. And it may have been the surviving one back then; not because it was better. Maybe the ones that run slower survived. I mean you know again not necessarily better but the surviving ones are basically the ones that then are shaped for any kind of subsequent evolutionary condition and environmental condition. 

But if you look at, for example, obesity. Obesity was selected for, basically the genes that predispose to obesity, were at a 2% frequency in Africa. They rose to 44% frequency in Europe.

{LF} Wow, that’s fascinating.

{MK}  Because you basically went through the Ice Ages and there was a scarcity of food, so you know there was a selection to being able to store every single calorie you consume. Eventually the environment changes. So the better allele, which was the fat storing allele, became the worst allele because it’s the fat storing allele. It still has the same function. So if you look at my genome, speaking of Mom calling, Mom gave me a bad copy of that gene, this FTO locus basically makes ….

{LF}   The one that has to do with obesity.

{MK}    Obesity. Yeah, basically now I have a bad copy from mom that makes me more likely to be obese. And I also have a bad copy from dad that makes me more likely to be obese, so I’m homozygous. And that’s the allele, it’s still the minor allele, but it’s at 44% frequency in Southeast Asia, 42% frequency in Europe, even though it started at 2% {in Africa}. It was an awesome allele to have a100 years ago. Right now, it’s a pretty terrible allele. 

So the other concept is that diversity matters. If we had 100 million nuclear physicists living on the earth right now we’d be in trouble. You need diversity. You need artists and you need musicians and you need mathematicians and you need, you know, politicians. Yes even those and you need like…. 

{LF}   Let’s not let’s not get crazy now. But because then if a virus comes along or whatever. 

{MK}    Exactly, exactly. So no, there’s two reasons. Number one, you want diversity in the immune repertoire and we have built-in diversity. So basically they are the most diverse …. basically if you look at our immune system there’s layers and layers of diversity. Like the way that you create your cells generates diversity because of the selection for the V(D)J  recombination that basically eventually leads to a huge number of repertoires but that’s only one small component of diversity. The blood type is another one.  The major histocompatibility complex, the HLA {Human leukocyte antigen} alleles are another source of diversity. So the immune system of humans is by nature incredibly diverse and that basically leads to resilience. So basically what I’m saying is that I don’t worry for the human species because we are so diverse immunologically we are likely to be very resilient against so many different attacks like this current virus. 

{LF}   So, you’re saying natural pandemics may not be something that you’re really afraid of because of the diversity in our genetic makeup. What about engineered pandemics? Do you have fears of us messing with the makeup of viruses or ….  Well, yeah let’s say with the makeup of viruses to create something that we can’t control and would be much more destructive that would come about naturally.

{MK}    Remember how we were talking about how smart evolution is? Humans are much dumber.

{LF} You mean like human scientists, human engineers?

{MK}  Humans, humans just like us.

{LF} Humans overall.

{MK}  But I mean even, you know, the sort of synthetic biologists. You know, basically if you were to create, you know, a virus like SARS that will kill other people you would probably start with SARS. So whoever, you know, would like to design such a thing would basically start with a SARS tree or at least some relative of SARS. The source genome for the current virus was something completely different. It was something that has never infected humans. No one in their right mind would have started there.

{LF}   But when you say sources like the nearest….

{MK}    The nearest relative is in a whole other branch, no species of which has ever infected humans in that branch. So you know, let’s put this to rest. This was not designed by someone to kill off the human race. 

{LF}   You don’t believe it was engineered

{MK}    The path to engineering a deadly virus would not come from this strain, that guy that was used. Moreover, there’s been various claims of: “Ha ha this was mixed and matched in the lab,” because the S1 protein has three different components, each of which has a different evolutionary tree. 

So you know, a lot of the popular press basically said: Aha, this came from pangolin and this came from, you know, all kinds of other species.” This is what has been happening throughout the coronavirus strain. So basically the S1 protein has been recombining across species all the time. 

Remember when I was talking about the positive-strand and negative-strands (sub-genomic RNAs) these can actually recombine and if you have two different viruses infecting the same cell, they can actually mix and match between the positive-strand and the negative-strand and basically create a new hybrid virus with recombination that now has the S1 from one and the rest of the genome from another. And this is something that happens a lot in S1 in ORF8, etc.  And that’s something that’s true of the whole tree.

{LF}   For the whole family of coronaviruses.

{MK}    Exactly. So it’s not like someone has been messing with this for millions of years and you know changing….

{LF} This happens naturally, Again, that’s beautiful. That somehow happens that they recombine, the two different strands can affect the body and recombine. So all of this magic actually happens inside hosts. Like all…..

{MK}    Yeah, that’s why they…. that’s why classification-wise, virus is not thought to be alive because it doesn’t self-replicate. It’s not autonomous, it’s something that enters a living cell and then co-opted to basically make it its own. But by itself, people ask me, how do we kill this bastard? Like you stop it from replicating. It’s not like a bacterium that will just live in a, you know, a puddle or something. It’s a virus, viruses don’t live without their hosts and they only live with their host for very little time. So if you stop it from replicating it’ll stop from spreading. I mean it’s not like HIV which can stay dormant for a long time, basically coronaviruses just don’t do that. They’re not integrating genomes, they are RNA genomes. So if it’s not expressed, it degrades. RNA degrades, it doesn’t just stick around.

{LF}   Well, let me ask, also about the immune system you mentioned. A lot of people kind of ask, you know: “How can we strengthen the immune system to respond to this particular virus or to the viruses in general?” Do you have, from a biological perspective thoughts, on what we can do as humans to strengthen our immune systems?

{MK}     If you look at death rates across different countries, people with less vaccination have been dying more. If you look at North Italy, the vaccination rates are abysmal there and a lot of people have been dying. If you look at Greece: very good vaccination rates, almost no one has been dying. So yes, there is a policy component. So Italy reacted very slowly. Greece reacted very fast. So you have many fewer people died in Greece but there might actually be a component of genetic immune repertoire. Basically, how did people die off, you know, in the history of the Greek population versus the Italian population, there’s…..

{LF}   That’s interesting to think about.

{MK}    And then there’s a component of what vaccinations did you have as a kid and what are the off-target effects of those vaccinations? So basically a vaccination can have two components. One is training your immune system against that specific insult. The second one is boosting up your immune system for all kinds of other things.

If you look at allergies, Northern Europe, super clean environments, tons of allergies. Southern Europe, my kids grew up eating dirt, no allergies. So growing up, I never had even heard of what allergies are. I was like really, allergies? And the reason is that I was playing in the garden, I was putting all kinds of stuff in my mouth from you know, all kinds of dirt and stuff. Tons of viruses there in terms of bacteria there, you know, my immune system was built up. So the more you protect your immune system from exposure, the less opportunity has to learn about non-self repertoire in a way that prepares it for the next insult.

{LF}  So, it’s a horizontal thing too, like throughout your lifetime and the lifetime of the people that  …. your ancestors, that kind of thing. What about …. So again it returns again to free will. On the free will side of things, is there something we can do to strengthen our immune systems in 2020? Is there like, you know, exercise diet, all that kind of stuff?

{MK}    So it’s kind of funny, there’s a cartoon that basically shows two windows with a teller in each window. One has a humongous line and the other one has no one, the one that has no one {the sign} above says health. No, it says exercise and diet and the other one says pill, there’s a huge line for pill. 

So we’re looking basically for magic bullets for sort of ways that we can, you know, beat cancer and beat coronavirus and beat this and beat that. It turns out that the window with just diet and exercise is the best way to boost every aspect of your health. If you look at Alzheimer’s: exercise and nutrition. I mean you’re like: “really, for my brain neurodegeneration.” Absolutely. If you look at cancer: exercise and nutrition. If you look at coronavirus: exercise and nutrition, every single aspect of human health gets improved. And one of the studies we’re doing now is basically looking at what are the effects of diet and exercise? How similar are they to each other? Were basically taken diet intervention and exercise intervention in human and in mice and we’re basically doing single cell profiling of a bunch of different tissues to basically understand how are the cells, both the stromal cells and the immune cells of each of these tissues responding to the effect of exercise. What are the communication networks between different cells where with the muscle that exercises sends signals through the bloodstream, through the lymphatic system, through all kinds of other systems that give signals to other cells that I have exercised and you should change in this particular way, which basically reconfigure those receptor cells with the effect of exercise.

{LF}  How well understood is the…. are those reconfigurations?

{MK}    Very little. We’re just starting now basically

{LF}   is the hope thereto understand the effect on…. So like the effect on the immune system?

{MK}    On the immune system, the effect on the brain, the effect on your liver, on your digestive system, on your adipocytes. Adipose, you know, is the most misunderstood organ. Everybody thinks fat is terrible. No fat is awesome. Your fat cells is what’s keeping you alive. Because if you didn’t have your fat cells, all those lipids and all those calories would be floating around in your blood and you’d be dead by now. Your adipocytes are your best friends and are basically storing all these excess calories so that they don’t hurt all of the rest of the body. And they’re also fat burning in many ways. 

So again, when you don’t have the homozygous version that I have, your cells are able to burn calories much more easily by sort of flipping a master metabolic switch that involves this FTO locus that I mentioned earlier and its target genes are  IRX3 and  IRX5. {[2]} They basically switch your adipocytes during their three first days of differentiation as they’re becoming mature adipocytes to basically become either fat burning or fat storing fat cells. 

And the fat burning fat cells {Brown adipose tissue} are your best friends. They’re much closer to muscle than they are to white adipocytes.

{LF}   Is there a lot of difference between people like that you could give ….. Science could eventually give advice that is very generalizable or is our differences in our genetic makeup like you mentioned, is that going to be basically something we have to be very specialized to individuals, any advice would give in terms of diet like we were just talking about.

{MK}   

Believe it or not, the most personalized advice that you give for nutrition don’t have to do with your genome. They have to do with your gut microbiome with the bacteria that live inside you. So most of your digestion is actually happening by species that are not human inside you. You have more non-human cells then you have human cells. You’re basically a giant bag of bacteria with a few human cells along. {laughs}

{LF}   And those do not necessarily have to do with your genetic makeup?

{MK}    They interact with your genetic makeup. They interact with your epigenome, they interact with your nutrition, they interact with your environment. They’re, you know, basically an additional source of variation. So when you’re thinking about sort of personalized nutritional advice, part of that is actually how do you match your microbiome? And part of that is how do we match your genetics? But again, you know, this is a very diverse set of, you know, contributors and the effect sizes are not enormous. So I think the science for that is not fully developed yet.

{LF}   Speaking of diet because I’ve wrestled in combat sports; sports my whole life where weight matters so you have to cut { lose weight rapidly} and all that stuff.One thing I’ve learned a lot about my body and it seems to be, I think true about other people’s bodies is that you can adjust a lot of things. That’s the miraculous thing about this biological system is like I fast often. I used to eat like 5-6 times a day and thought that was absolutely necessary. How could you not eat that often? And then when I started fasting, your body adjusts to that and you learn how to not eat, you know, and it’s as if you just give it a chance for a few weeks…. actually over a period of a few weeks, your body can adjust to anything. And that’s such a beautiful thing.

{MK}    So I’m a computer scientist and I’ve basically gone through periods of 24 hours without eating or stopping and you know, then I’m like, oh, must eat and I eat a ton. I used to order two pizzas just with my brother and you know, like I’ve gone through these extremes as well and I’ve gone the whole intermittent fasting thing. So I can sympathize with you both on the seven meals a day to the zero meals a day. 

So I think when I say everything in moderation, I actually think your body responds interestingly to these different changes in diet. I think part of the reason why we lose weight with pretty much every kind of change in behavior is because our epigenome and the set of proteins and enzymes that are expressed and our microbiome are not well suited to that nutritional source. And therefore they will not be able to sort of catch everything that you give them. And then, you know, a lot of that will go undigested. And that basically means that your body can then, you know, lose weight in the short term, but very quickly will adjust to the new normal. And then we’ll be able to sort of perhaps gain a lot of weight from the diet. 

So anyway, I mean there’s also studies in factories where basically people, you know, dim the lights. And then suddenly everybody started working better. It was like: “Wow, that’s amazing.”

Three weeks later they made the lights a little brighter, everybody started working better. {Hawthorne effect} So any kind of intervention has a placebo effect of: “Wow, now I’m healthier and I’m going to be running more often,” etc. So it’s very hard to uncouple the placebo effect of “Wow, I’m doing something to intervene on my diet,” from the: “Wow, this is actually the right thing for me.” So you know….

{LF}   Yeah. From the perspective from a nutrition science, psychology, both things I’m interested, especially psychology, it seems that it’s extremely difficult to do good science because there’s so many variables involved, it’s so difficult to control the variables, so difficult to do sufficiently large scale experiments, both sort of in terms of number of subjects and temporal, like how long you do the study for; that it just seems like it’s not even a real science for now, like nutrition science.

{MK}    I want to jump into the whole placebo effect for a little bit here. And basically, talk about the implications of that. If I give you a sugar pill and tell you it’s a  sugar pill, you won’t get any better. But if I tell you  {it’s a} sugar pill and tell you, and I’ll tell you: “Wow, this is an amazing drug. It actually will stop your cancer;” your cancer will actually stop with much higher probability. What does that mean? 

{LF} That’s so amazing.

{MK}  That means that if I can trick your brain into thinking that I’m healing you, your brain will basically figure out a way to heal itself, to heal the body. And that tells us that there’s so much that we don’t understand in the interplay between cognition and our biology, that if we were able to better harvest the power of our brain to sort of impact the body through the placebo effect, we would be so much better in so many different things. Just by tricking yourself into thinking that you’re doing better, you’re actually doing better. So there’s something to be said about positive thinking, about optimism, about sort of, you know, just getting your brain and your mind into the right mindset that helps your body and helps your entire biology.

{LF}   Yeah, from a science perspective that’s just fascinating. Obviously, most things about the brain is a total mystery for now, but that’s a fascinating interplay that the brain can reduce…. the brain can help cure cancer. I don’t even know what to do with that.

{MK}    I mean, the way to think about that is the following the converse of the equation is something that we are much more comfortable with, like, oh, if you’re stressed than your heart, rate might rise and all kinds of sort of toxins might be released and that can have a detrimental effect on your body etc, etc. So maybe it’s easier to understand your body healing from your mind by your mind is not killing your body or at least it’s killing it less. So I think that, you know, that aspect of the stress equation is a little easier for most of us to conceptualize. But then the healing part is you know, perhaps the same pathways, perhaps different pathways. But again, something that is totally untapped scientifically.

{LF}   I think we tried to bring this question up a couple of times but let’s return to it again. What do you think is the difference between the way a computer represents information {and} the human genome represents and stores information. Like what…. and maybe broadly what is the difference between how you think about computers and how you think about biological systems?

{MK}    So I made a very provocative claim earlier that we are a digital computer, like that at the core lies a digital code. And that’s true in many ways. But surrounding that digital core there’s a huge amount of analog. If you look at our brain, it’s not really digital. If you look at our sort of RNA and all of that stuff inside ourselves, it’s not really digital. It’s really analog in many ways. But let’s start with the code and then we’ll expand to the rest. 

So the code itself is digital. So there’s genes. You can think of the genes as I don’t know, the procedures, the functions inside your language and then somehow you have to turn these functions on. How do you call a gene? How do you call that function?

The way that you would do it in old programming languages is goto address whatever in your memory and then you start running from there. And you know modern programming languages have encapsulated this into functions and objects and all of that and it’s nice and cute. But in the end, deep down, there’s still an assembly code that says go to that instruction and it runs that instruction. 

If you look at the human genome and the genome of pretty much most species out there it’s There’s no goto function. You just don’t start in, you know, transcribing in position 1305; you know 13,527 in chromosome 12. You instead have content based indexing. 

So at every location in the genome in front of the genes that need to be turned on, I don’t know when you drink coffee, there’s a little coffee marker in front of all of them. And whenever your cells that metabolize coffee need to metabolize coffee, they basically see coffee and they’re like:  “Ooh let’s go turn on all the coffee marked genes.” So there’s basically these small motifs, these small sequences that we call regulatory motifs. They’re like patterns of DNA. They’re only eight characters long or so like GATGCA etc. 

And these motifs work in combinations and every one of them has some recruitment affinity for a different protein that will then come and bind it. And together collections of these motifs create regions that we call regulatory regions that can be either promoters near the beginning of the gene and that basically tells you where the function actually starts where you call it. And then enhancers that are looping around of the DNA that basically bring the machinery that binds those enhancers and then bring it onto the promoter which then recruits the right sort of ribosome and the polymerase and all of that thing which will first transcribe and then export and then eventually translating in the cytoplasm you know whatever RNA molecule. 

So the beauty of the way that the digital computer that’s the genome works is that it’s extremely fault tolerant. If I took your hard drive and I messed with 20% of the letters in it, of the zeros and ones and flipped them you’d be in trouble. If I take the genome and I flipped 20% of the letters, you probably won’t even notice. And that resilience….

{LF}   That’s fascinating.

{MK}    is a key design principle. And again anthropomorphizing here. But it’s a key driving principle of how biological systems work. They are first resilient and then anything else. And when you look at this incredible beauty of life from the most, I don’t know, beautiful, I don’t know, human genome maybe of humanity and all of the ideas that you come with it; to the most terrifying genome, like I don’t know, COVID-19, SARS-CoV-2, and the current pandemic. You basically see this elegance as the epitome of clean design. But it’s dirty. It’s a mess. It’s you know, the way to get there is hugely messy. And that’s something that we as computer scientists don’t embrace. We like to have clean code. You know, like in engineering they teach you about compartmentalization about separating functions, about modularity, about hierarchical design. None of that applies in biology.

{LF}  Testing.

{MK}  Testing. Sure, biology does plenty of that. But I mean through evolutionary exploration but if you look at biological systems first they are robust and then they specialize to become anything else. And if you look at viruses, the reason why they’re so elegant. When you look at the design of this genome, it seems so elegant. And the reason for that is that it’s been stripped down from something much larger because of the pressure to keep it compact. So many compact genomes out there have ancestors that were much larger. You don’t start small and become big. You go through a loop of: add a bunch of stuff, increase complexity and then you know, slim it down. 

And one of my early papers was in fact on genome duplication.[3]  One of the things we found is that baker’s yeast, which is the, you know, the yeast that you used to make bread but also the yeast that used to make wine. Which is basically the dominant species when you go in the field of Tuscany and you say, you know what’s out there? It’s basically Saccharomyces cerevisiae. Or the way my Italian friends say: Saccharomyces cerevisiae {with an Italian pronunciation}

{LF}   Which means what?

{MK}    Oh okay. I’m sorry. I’m Greek. So yeah, Saccharomyces. “Saccharo” is sugar “myces” is fungus. Yes. “cerevisiae”,  cerveza beer. It means the sugar fungus of beer. You know, less good sounding to the ear.

{LF}   Still poetic.

{MK}    So anyway Saccharomyces cerevisiae, basically is the major bakers yeast out there. It’s the descendant of a whole genome duplication. Why would a whole genome duplication even happen? When it happened is coinciding with about 100 million years ago and the emergence of fruit bearing plants? Why fruit bearing plants? Because animals would eat the fruit, would walk around and poop huge amounts of nutrients along with the seed for the plants to spread. Before that plants were not spreading through animals. They were spreading through wind and all kinds of other ways. But basically the moment you have fruit bearing plants, these plants are basically creating this abundance of sugar in the environment. So there’s an evolutionary niche that gets created and in that evolutionary niche you basically have enough sugar that whole genome duplication which initially is a very massive event allows you to then, you know, relieve some of that complexity.

{LF}   So I have to pause. What does genome duplication mean?

{MK}    That basically means that instead of having eight chromosomes, you’re going to now have sixteen chromosomes.

{LF}   So but the duplication at first when you have six….  when you go to 16 you’re not using that.

{MK}    Oh yeah you are. Yeah. So basically from one day to the next you went from having eight chromosomes to having 16 chromosomes. Probably a nondisjunction event during the duplication during a division. So you basically divide the cell instead of half the genome going this way and half the genome going the other way after duplication of the genome, you basically have all of it going to one cell. And then there’s a sufficient messiness there that you end up with slight differences that make most of these chromosomes be actually preserved. It’s a long story short to me

LF: But it’s a big upgrade, right? So that’s ….

{MK}    Not necessarily because what happens immediately thereafter is that you start massively losing tons of those duplicated genes. So 90% of those genes were actually lost very rapidly after whole gene duplication. And the reason for that is that biology is not intelligent, it is just ruthless selection, random mutation. So ruthless selection basically means that as soon as one of the random mutations hits one gene, ruthless selection just kills off that gene. 

It’s just you know, if you have a pressure to maintain a small compact genome you will very rapidly lose the second copy of most of your genes and a small number (10%) were kept in two copies. And those have to do a lot with environment adaptation, with the speed of replication, with the speed of translation and with sugar processing. So I’m making a long story short to basically say that evolution is messy. 

The only way like…. so the example that I was giving of messing with 20% of your bits in your computer; totally bogus. Duplicating all your functions and just throwing them out there in the same, you know, function; just totally bogus. Like this will never work in an engineered system but biological systems because of this content based indexing and because of this modularity that comes from the fact that the gene is controlled by a series of tags. And now if you need this gene in another setting you just add some more tags that will basically turn it on also in those settings. So this gene is now pressured to do two different functions and it builds up complexity. I see whole gene duplication and gene duplication in general as a way to relieve that complexity. 

So you have this gradual build up of complexity as functions get sort of added onto the existing genes and then boom you duplicate your workforce and you now have two copies of this gene.

One will probably specialize to do one and the other one will specialize to do the other or one will maintain the ancestral function, the other one will sort of be free to evolve and specialize while losing the ancestral functions and so on and so forth. So that’s how genomes evolve. They’re just messy things but they’re extremely fault tolerant and they’re extremely able to deal with mutations because that’s the very way that you generate new functions. 

So new functionalization comes from the very thing that breaks it. So even in the current pandemic many people are asking me which mutations matter the most and what I tell them is: “Well, we can study the evolutionary dynamics of the current genome to then understand which mutations have previously happened or not and which mutations happen in genes that evolved rapidly or not.” And one of the things we found for example is that the genes that evolved rapidly in the past are still evolving rapidly now in the current pandemic. The genes have evolved slowly in the past are still evolving slowly.

{LF}   Which means that they’re useful.

{MK}    Which means that they are under the same evolutionary pressures. But then the question is what happens in specific mutations? So if you look at the D614G mutation that’s been all over the news. So in position 614 in the amino acids of the S protein there’s a D {aspartic acid} to G {glycine} mutation that sort of has creeped over the population. That mutation we found out through my work disrupts a perfectly conserved nucleotide position. That has never been changed in the history of millions of years of equivalent mammalian evolution of these viruses. That basically means that it’s a completely new adaptation to human. And that mutation has now gone from 1% frequency to 90% frequency in almost all outbreaks.

{LF}   So there’s a mutation. I like how you say the 416, that was okay, 614….

{MK}    D614G

{LF}   Literally, so what you’re saying is that this is like a chess move. So it just mutated one letter to another and that hasn’t happened before. And this somehow….  this mutation is really useful.

{MK}    It’s really useful in the current environment of the genome which is moving from human to human. When it was moving from bat to bat, it couldn’t care less for that mutation but its environment specific. So now that is moving from human to human who is moving way better, like by orders of magnitude.

{LF}   Okay, so you’re like tracking this evolutionary dynamics, which is fascinating. But what do you do with that? So what does that mean? What does this mean? What do you make, what do you make of this mutation in trying to anticipate I guess is the…. is one of the things you’re trying to do is anticipate where……  how this unrolls into the future, this evolutionary dynamics?

{MK}    Such a great question. So there’s two things, remember when I was saying earlier mutation is the path to new things, but also the path to break all things. So what we know is that this position was extremely preserved through gazillions of mutations; that mutation was never tolerated when it was moving from bats to bats. So that basically means that that position is extremely important in the function of that protein. That’s the first thing it tells. The second one is that that position was very well suited to bat transmission but now is not well suited to human transmission. So it got rid of it and it now has a new version of that amino acid that basically makes it much easier to transmit from human to human. 

So in terms of the evolutionary history teaching us about the future, it basically tells us here is the regions that are currently mutating. Here’s the regions that are most likely to mutate going forward. As you’re building a vaccine, here’s what you should be focusing on in terms of the most stable regions that are the least likely to mutate or here’s the newly evolved functions that are most likely to be important because they’ve overcome these local maximum that it had reached in the bat transmission. So anyway, it’s a tangent to basically say that evolution works in messy ways. And the thing that you would break is the thing that actually allows you to first go through a lull and then reach a new local maximum. And I often like to say that if engineers had basically designed evolution, we would still be perfectly replicating bacteria. Because it’s by making the bacterium worse that you allow evolution to reach a new optimum.

{LF}   Just a pause on that. That’s so profound. That’s so profound for the entirety of these scientific and engineering disciplines.

{MK}    Exactly. We as engineers need to embrace breaking things. We, as engineers, need to embrace robustness as the first principle beyond perfection because nothing’s going to ever be perfect. And when you’re sending a satellite to Mars, when something goes wrong, it’ll break down as opposed to building systems that tolerate failure and our resilience to that, and in fact get better through that.

{LF}   So the SpaceX approach versus NASA for the {laughs}

{MK}  For example.

{LF} Is there something we can learn about the incredible…. take lessons from the incredible biological systems in their resilience in their… in the mushiness, the messiness to our computing systems to our computers.

{MK}    It would basically be starting from scratch in many ways. It would basically be building new paradigms that don’t try to get the right answer all the time, but try to get the right answer most of the time or a lot of the time. 

{LF}   Do you see deep learning systems in the whole world of machine learning is kind of taking a step in that direction?

{MK}    Absolutely. Absolutely. Basically, by allowing much more natural evolution of these parameters. You basically and …. and if you look at sort of deep learning systems again, they’re not inspired by the genome aspect of biology are inspired by the brain aspect of biology. And again, I want you to pause for a second and realize the complexity of the entire human brain with trillions of connections within our you know, neurons; with millions of cells talking to each other is still encoded within that same genome. That same genome encodes every single freaking cell type of the entire body. Every single cell is encoded by the same code. 

And yet specialization allows you to have this single viral-like genome that self-replicates the single module, modular automaton, work with other copies of itself. It’s mind boggling. Create complex organs through which blood flows. And what is that blood? The same freaking genome. Create organs that communicate with each other. And what are these organs? The exact same genome. Create a brain that is innervated by massive amounts of blood pumping energy to it, 20% of our energetic needs, to the brain from the same genome. And all of the neuronal connections, all of the auxiliary cells, all of the immune cells. The astrocytes, the oligodendrocytes, the neurons, the excitatory {neurons}, the inhibitory neurons, all of the different classes of pericytes, the blood-brain barrier, all of that: same genome.

{LF}   One way to see that in a sad ….. this one is beautiful. The sad thing is thinking about the trillions of organisms that died to create that.

{MK}    You mean on the evolutionary path?

{LF}   On the evolutionary path of humans. It’s crazy. These two descendants of apes are just talking on the podcast. Okay, so mind boggling,

{MK}    Just to boggle our minds a little bit more, us talking to each other. We are basically generating a series of vocal utterances through our pulsating of vocal cords received through this. {Pointing to his ear}  The people who listen to this are taking a completely different path to that information transfer yet through language. 

But imagine if we could connect these brains directly to each other. The amount of information that I’m condensing into a small number of words is a huge funnel which then you receive and you expand into a huge number of thoughts from that small funnel. 

In many ways, engineers would love to have the whole information transfer. Just take the whole set of neurons and throw them away. I mean throw them to the other person. This might actually not be better because in your misinterpretation of every word that I’m saying, you are creating a new interpretation that might actually be way better than what I meant in the first place. The ambiguity of language perhaps might be the secret to creativity. 

Every single time you work on a project by yourself, You only bounce ideas with one person and your neurons are basically fully cognizant of what these ideas are . At the moment you interact with another person, the misinterpretations that happen might be the most creative part of the process with my students every time we have a research meeting, I very often pause and say, let me repeat what you just said in a different way, and I sort of go on and brainstorm with what they were saying, but by the third time it’s not what they were saying at all, and when they pick up what I’m saying, like, oh well, da da da, now they have sort of, learned something very different from what I was saying. And that is the same kind of messiness that I’m describing in the genome itself. It’s sort of embracing the messiness

{LF}   And that’s a feature, not a bug.

{MK}    Exactly. And in the same way, when you’re thinking about these deep learning systems that will allow us to sort of be more creative perhaps, or learn better approximations of these complex functions, again, tuned to the universe that we inhabit. You have to embrace the breaking, you have to embrace the how do we get out of these local optima? And a lot of the design paradigms that have made deep learning so successful are ways to get away from that, ways to get better training by sort of sending long range messages, the LSTM models {Long short-term memory} and the sort of feedforward loops that, you know, sort of jumped through layers of convolutional neural network, all of these things are basically ways to push you out of this local maxima, and that’s what evolution does, that’s what language does, that’s what conversation and brainstorming does, that’s what our brain does? So, you know, this design paradigm is something that’s pervasive and yet not taught in schools, not taught in engineering schools. where everything’s minutely modular wrist to make sure that we never deviate from, you know, whatever signal we’re trying to admit as opposed to let all hell breaks loose because that’s the that’s the path to paradise. 

{LF}   The path to paradise. Yeah, I mean, it’s difficult to know how to teach that and what to do with it. I mean, it’s difficult to know how to build up the scientific method around messiness.

{MK}    I mean it’s not all messiness. We need some cleanliness. And going back to the example with Mars, that’s probably the place where I want to sort of moderate error as much as possible and sort of control the environment as much as possible. But if you’re trying to repopulate Mars, well, maybe messing is a good thing then.

{LF}   On that, you quickly mentioned this in terms of us using our vocal cords to speak on a podcast. Elon Musk and Neuralink are working on trying to plug, as per our discussion with computers and biological systems, to connect it to….he’s trying to connect our brain to a computer to create a brain computer interface where they can communicate back and forth on this line of thinking. Do you think this is possible,  to bridge the gap between our engineered computing systems and the messy biological systems?

{MK}  

My answer would be absolutely, you know, there is no doubt that we can understand more and more about what goes on in the brain and we can sort of train the brain. I don’t know if you remember the Palm Pilot?

{LF}   Yeah, Palm Pilot.

{MK}    Remember this whole sort of alphabet {Graffiti for Palm OS} that they had created, am thinking of the same thing? It’s basically you had, you had a little pen and for every character you had a little scribble that was unique that the machine could understand and that instead of trying the machine trying to teach the machine to recognize human characters, you have basically they figured out that it’s better and easier to train humans to create human-like characters that the machine is better at recognizing. 

So, in the same way I think what will happen is that humans will be trained to be able to create the mind pattern that the machine will respond to before the machine truly comprehends your thoughts. So the first human-brain interfaces will be tricking humans to speak the machine language where with the right set of electrodes, I can sort of trick my brain into doing this. And this is the same way that many people teach….  learned to control artificial limbs. You basically try a bunch of stuff and eventually you figure out how your limbs work. 

That might not be very different from how humans learn to use their natural limbs when they first grew up basically you have these, you know, neoteny period of you know, this puddle of soup inside your brain, trying to figure out how to even make neuronal connections before you’re born and then learning sounds in utero of, you know, all kinds of echoes and you know, eventually getting out in the real world. And I don’t know if you’ve seen newborns but they just stare around a lot, you know, one way to think about this as a machine learning person is oh they’re just training their edge detectors and eventually they figure out how to train their edge detectors. They work through the second layer of the visual cortex and the third layer and so on and so forth. 

And you basically have this learning how to control your limbs. That probably comes at the same time, you’re sort of throwing random things there and you realize that, wow, when I do this thing, my limb moves. 

Let’s do the following experiment, take a breath. What muscles did you flex, now? Take another breath and think what muscles reflex. The first thing that you’re thinking when you’re taking a breath is the impact that he has in your lungs. You’re like, oh I’m now going to increase my lung, so I’m not going to bring air in, but what you’re actually doing is just changing your diaphragm. That’s not conscious, of course. You never think of the diaphragm as a thing and why is that? That’s probably the same reason why I think of moving my finger when I actually move my finger, I think of the effect. Instead of actually thinking of whatever muscles twitching that actually causes my finger to move. 

So we basically, in our first years of life, build up this massive look up table between whatever neuronal firing we do and whatever action happens in our body that we control. If you have a kid grow up with a third limb, I’m sure they’ll figure out how to control them, probably at the same rate as their natural limbs.

{LF}   And a lot of the work would be done by the …. so if a third limb is a computer, you kind of have not a faith, but a thought that the brain might be able to figure out….

{MK}  Absolutely

{LF} Like the plasticity would come from the brain, like the brain would be cleverer than the machine at first.

{MK}    When I talk about a third limb, that’s exactly what I’m saying is an artificial limb that basically just controls your mouse while you’re typing, you know, perfectly natural thing. I mean, again, you know, in a few hundred years.

{LF}  Maybe sooner than that.

{MK}   But basically there’s….  as long as the machine is consistent in the way that it will respond to your brain impulses, you’ll figure out how to control that, and you could play tennis with your third limb.

And let me go back to consistency. People who have dramatic accidents that basically take out a whole chunk of their brain can be taught to co-opt other parts of the brain to then control that part. You can basically build up that tissue again and eventually train your body how to walk again and how to read again and how to play again and how to think again, how to speak a language again, etc. 

So there’s a massive amount of malleability that happens, you know, naturally in our way of controlling our body or brain or thoughts or vocal cords or limbs, et cetera. And human-machine interfaces are not inevitable if we figure out how to read these electric impulses. But the resolution at which we can understand human thought right now is nil, is ridiculous. 

So how are human thoughts encoded? It’s basically combinations of neurons that co-fire and these create these things called engrams that eventually form memories and so and so forth. We know nothing of all that stuff. So before we can actually read into your brain that you want to build a program that does this and this and that we need a lot of neuroscience?

{LF}   Well, so to push back on that, do you think it’s possible that without understanding the functionally about the brain or from neuroscience or cognitive science or psychology? Whichever level of the brain will look at. Do you think we just connect, connect them just like per your previous point? If we just have a high enough resolution between …. connection between Wikipedia and your brain; the brain will just figure it out with less understanding because that’s one of the innovations of Neuralink is they’re increasing the number of connections to the brain to like several thousand which before was, you know, in the dozens or whatever.

{MK}    You’re still off by a few orders of magnitude, on the order of seven.

{LF}   Right. But the thing is, the hope is, if you increase that number more and more and more, maybe you don’t need to understand anything about the actual ….. how human thought is represented in the brain. You can just let it figure it out. 

{MK}    {Sentence I couldn’t decipher} I know. 

{LF} Yeah, exactly. Yeah. 

{MK}  So yeah, sure.

{LF}   You don’t have faith in the plasticity of the brain to that degree.

{MK}    It’s not about brain plasticity, it’s about the input aspect. Basically. I think on the output aspect, being able to control the machine is something that you can probably train your neural impulses that you’re sending out to sort of match whatever response you see in the environment. If this thing {points to microphone} moved every single time I thought a particular thought, then I could figure out, I could hack my way into moving this thing with just a series of thoughts. I could think: “Guitar, Piano, tennis ball” and then this thing would be moving and then you know, I would just have the series of thoughts that would sort of result in the impulses that will move this thing the way that I wanted and then eventually it’ll become natural because I won’t even think about it. I mean in the same way that we control our limbs in a very natural way, but babies don’t do that babies have to figure it out and you know, some of it is hard coded but some of them is actually learned based on the whatever soup of neurons you ended up with, whatever connections you pruned them to and eventually you were born with, you know, a lot of that is coding in the genome but a huge chunk of that is stochastic. And sort of the way that you sort of create all these neurons, they migrate, they form connections, they sort of spread out. They have particular branching patterns. But then the connectivity itself, unique in every single new person. 

All this to say that on the output side: Absolutely. I’m very, very you know hopeful that we can have machines that read thousands of these neuronal connections on the output side, but on the input side: Oh boy,I don’t expect any time in the near future we’ll be able to sort of send a series of impulses that will tell me, oh Earth to some distance 7.5 million ….  etc, like nowhere. I mean I think language will still be the input way rather than sort of any kind of more complex.

{LF}   It’s a really interesting notion that the ambiguity of languages is a feature and we evolved for millions of years to take advantage of that ambiguity. 

{MK}    Exactly. And yet no one teaches us the subtle differences between words that are near cognates and yet evokes so much more than, you know, one from the other. And yet, you know, when you’re choosing words from a list of 20 synonyms, you know exactly the connotation of every single one of them and that’s something that, you know, is there. So, yes, there’s ambiguity, but there’s all kinds of connotations and in the way that we select our words, we have so much baggage that we’re sending along the way that were emoting the way that we’re moving our hands every single time we speak the, you know, the pauses, the eye contact, etc, so much higher baud rate than just a vocal, you know, a string of characters. 

{LF}   Well, let me just take a small tangent on that.

{MK}    Oh, tangent, we haven’t done that yet ….

{LF}   We’ll turn to the origin of life.

{LF}   So I mean you’re Greek but I’m going on this personal journey. I’m going to Paris for the explicit purpose of talking to one of the most famous….  a couple who is a famous translators of Russian literature, Dostoevsky, Tolstoy and they go, that’s their art is the translation. And everything I’ve learned about the translation art, it makes me feel …. it’s so profound in a way that’s so much more profound than the natural language processing papers I read in the machine learning community.Tthat there’s such depth to language thatI don’t know what to do with, I don’t know if you’ve experienced that in your own life with knowing multiple languages. I don’t know what to do, I don’t know how to make sense of it, but there’s so much loss in translation between Russian and English and getting a sense of that. 

Like, for example, there’s like just taking a single sentence from Dostoevsky and like there’s a lot of them, you could, you could talk for hours about how to translate that sentence properly: that captures the meaning, the period, the culture, the humor, the wit, the suffering, that was in the context of the time. All of that could be a single sentence.you could, you could talk forever about what it takes to translate that correctly. I don’t know what to do with that.

{MK}    So, being Greek, it’s very hard for me to think of a sentence or even a word without going into the full etymology of that word. Breaking up every single atom of that sentence and every single atom of these words and rebuilding it back up. 

I have three kids and the way that I teach them Greek is the same way that, you know, the documentary I was mentioning earlier about sort of understanding the deep roots of all of these, you know, words. And it’s very interesting that every single time I hear a new word that I’ve never heard before, I go and figure out the etymology of that word because I will never appreciate that word without understanding how it was initially formed.

{LF}   Interesting. But how does that help? Because that’s not the full picture.

{MK}    No, no, of course of course. But what I’m trying to say is that knowing the components teaches you about the context of the formation of that word and sort of the original usage of that word. And then of course the word takes new meaning as you create it from its parts and that meaning then gets augmented and two synonyms that sort of have different roots will actually have implications that carry a lot of that baggage of the historical provenance of these words. So before working on genome evolution, my passion was evolution of language and sort of tracing cognates across different languages through their etymologies.

{LF}   And that’s fascinating that there’s parallels between ….

{MK}  Of course. 

{LF} I mean the idea that there’s evolutionary dynamics to our language.

{MK}    Yeah, every single word that you utter: parallels, parallels. What does parallels mean: “Para” means side by side; “allel” from alleles which means identical twins: parallel. I mean name any word and there’s so much baggage, so much beauty in how that word came to be and how this word took a new meaning than the sum of its parts.

{LF}   Yeah. And they’re just words. They don’t have any physical ….

{MK}    Exactly. And now you take…. they’re just words and you weave them into a sentence. The emotional invocations of that weaving are fathomless.

{LF}   And there are …. all of those emotions all live in our in the brains of humans

{MK}   In the eye of the beholder. No seriously, you have to embrace this concept of the eye of the beholder. It’s the conceptualization that nothing takes meaning with one person creating it. Everything takes meaning in the receiving end and the emergent properties of these communication networks where every single….  you know, if you look at the network of our cells and how they’re communicating with each other, every cell has its own code. This code is modulated by the epigenome. This creates a bunch of different cell types. Each cell type now has its own identity. Yet they all have the common root or the stem cells that led to them. Each of these identities is now communicating with each other. They take meaning in their interaction. There is an emergent property that comes from a bunch of cells being together that is not in any one of the parts.

If you look at neurons communicating again, these engrams don’t exist in any one neuron. They exist in the connection, in the combination of neurons. And the meaning of the words that I’m telling you is empty until it reaches you and it affects you in a very different way than it affects whoever’s here listening to this conversation now. Because of the emotional baggage that I’ve grown up with that you’ve grown up with and that they’ve grown up with. 

And that’s I think that’s the magic of translation. If you start thinking of translation as just simply capturing that emotional set of reactions that you evoke. You need a different set of words to evoke that same set of reactions to a French person than to a Russian person because of the baggage of the culture that we grew up in. 

{LF}   Yeah, I mean….

{MK}    So, so basically you shouldn’t find the best word. Sometimes it’s a completely different sentence structure that you will need matched to the cultural context of the target audience that you have.

{LF}   Yeah, it’s I mean you’re just,….  I usually don’t think about this. But right now there’s this feeling, it’s a reminder that it’s just you and I talking but there’s several hundred thousand people who will listen to this. There’s some guy in Russia right now running like in Moscow listening to us. There’s somebody in India I guarantee you there’s somebody in China and South America. There’s somebody in Texas and they all have different ….

{MK}    … emotional baggage.

{LF} They probably got angry earlier on about the whole discussion about Coronavirus and about some aspect of it, yeah and there’s that network effect. That’s ….

{MK}    It’s a beautiful thing and this lateral transfer of information, that’s what makes the collective quote unquote genome of humanity so unique from any other species.

{LF}   So you somehow miraculously wrapped it back to the very beginning of when we’re talking about the human …. the beauty of the human genome. So I think this is the right time unless we want to go for a six to eight hour conversation, we’re going have to talk again. But I think for now to wrap it up this, is the right time to talk about the biggest, most ridiculous question of all: meaning of life. Off mike, you mentioned to me that you had your 42nd birthday, 42nd being a very special, absurdly special number and you had a kind of get together with friends to discuss the meaning of life. So let me ask you in your…. as a biologist, as a computer scientist, and as a human, what is the meaning of life?

{MK}    I’ve been asking discretion for a long time ever since my 42nd birthday. But well before that and even planning the Meaning of Life symposium.  And symposium “sym” means together; “posi” actually means to drink together. So symposium is actually a drinking party. 

{LF}   So can you actually elaborate about this Meaning of Life symposium that you put together. It’s the most genius idea I ever heard.

{MK}    So 42 is obviously the answer to life, the universe and everything from the Hitchhiker’s Guide to the Galaxy. And as I was turning 42…. I’ve had the theme for every one of my birthdays. When I was turning 32 it’s 100,000 in binary. So I celebrated my 100,000 binary binary birthday and I had a theme of going back 100,000 years, you know, let’s dress something in the last 100,000 years anyway, was I’ve always had these, that’s

{LF}   You’re such an interesting human being. Okay, that’s awesome. 

{MK}    I’ve always had these sort of numerology related announcements for my birthday party. So, what came out of that Meaning of Life symposium is that I basically asked 42 of my colleagues, 42 of my friends, 42 of my, you know, collaborators to basically give seven minute speeches on the meaning of life, each from their perspective. 

And I really encourage you to go there { Meaning of Life Symposium Videos} because it’s mind boggling that every single person said a different answer. Every single person started with: ”I don’t know what the meaning of life is, but….”  and then give this beautifully eloquent answer and they were all different, but they all were consistent with each other and mutually synergistic and together, forming a beautiful view of what it means to be human in many ways. 

Some people talked about the loss of their loved one, their life partner for many, many years and how their life changed through that. Some people talked about the origin of life. Some people talked about the difference between purpose and meaning.

 I’ll you know maybe quote one of the answers which is this linguistics professor friend of mine at Harvard who basically said that she was going to, she’s Greek as well, and she said I will give a very Pythian answer. { See Calliopi Dourou – Meaning of Life – “Become One”: Excellence, Sharing, Renewal } So, Pythia was the Oracle of Delphi who would basically give these very cryptic answers very short but interpretable in many different ways. There was this whole set of priests who were tasked with interpreting what Pythia had said and very often you would not get a clean interpretation but she said I will be like Pythia and give you a very short and multiple interpreter will answer. But unlike her I will actually also give you three interpretations. And she said: “The answer to the meaning of life is become one.” And the first interpretation is like a child become one year old with the excitement of discovering everything about the world.

 Second interpretation: in whatever you take on become one, the first, the best, excel, drive yourself to perfection for every one of your tasks. And become one when people are separate, become one, come together, learn to understand each other. 

{LF}   Damn! That’s an answer.

{MK}    And one way to summarize this whole Meaning of Life symposium is that the very symposium was illustrating the quest for meaning, which might itself be the meaning of life. This constant quest for something sublime, something human, something intangible. Some, you know, aspect of what defines us as a species and as an individual. Both the quest of me as a person through my own life. 

But the meaning of life could also be the meaning of all of life. What is the whole point of life? Why life? Why life itself? Because we’ve been talking about the history and evolution of life, but we haven’t talked about why life in the first place, is life inevitable? Is life part of physics?  Does life transcend physics by fighting against entropy, by compartmentalizing and increasing concentrations rather than diluting away. 

Is life a distinct entity in the Universe beyond the traditional, very simple physical rules that govern gravity and electromagnetism and all of these forces? Is life another force? Is there a life force? Is there a unique kind of set of principles that emerge, of course, built on top of the hardware of physics? But is it sort of a new layer of software or a new layer of a computer system? 

So that’s at the level of, you know, big questions. There’s another aspect of gratitude. Basically what I, you know what I like to say is during this pandemic, I’ve basically worked from six AM until seven PM every single day nonstop, including Saturday and Sunday. I’ve basically broken all boundaries of where life, personal life begins and work life, you know, ends. And that has been exhilarating for me. Just just the intellectual pleasure that I get from a day of exhaustion where at the end of the day my brain is hurting. I’m telling my wife, wow, I was useful today. And there’s a certain pleasure that comes from feeling useful and there is certain pleasure that comes from feeling grateful. So I’ve written this little sort of prayer for my kids to say at bedtime every night where they basically say: Thank you God for all you have given me and give me the strength to give onto others with the same love that you have given onto me.”

 We as a species are so special. The only ones who worry about the meaning of life and maybe that’s what makes us human and what I like to say to my wife and to my students during this pandemic work extravaganza is every now and then they asked me, but how do you do this? And I’m like, I’m a workaholic. I love this. This is me in the most unfiltered way. The ability to do something useful, to feel that my brain is being used, to interact with the smartest people on the planet day in day out and to help them discover aspects of the human genome, of the human brain, of human disease and the human condition that no one has seen before with data that we’re capturing that has never been observed. 

And there’s another aspect which is on the personal life. Many people say, oh, I’m not going to have kids. Why bother? I can tell you as a father. They’re missing half the picture, if not the whole picture teaching my kids about my view of the world and watching through their eyes, the naivete with which they start and the sophistication with which they end up. The understanding that they have of not just the natural world around them, but of me too. The unfiltered criticism that you get from your own children that knows no bounds of honesty. And I’ve grown components of my heart that I didn’t know I had until you sense that fragility, that vulnerability of the children, that immense love and passion, the unfiltered egoism that we as adults learn how to hide so much better. It’s just this bag of emotions that tell me about the raw materials that make a human being and how these raw materials can be arranged with more sophistication that we learn through life to become truly human adults. 

But there’s something so beautiful about seeing that progression between them. The complexity of the language growing as more neural connections are formed. To realize that the hardware is getting rearranged as their  software is getting implemented on that hardware That their frontal cortex continues to grow for another 10 years. Their neuronal connections are continuing to form, new neurons that actually get replicated and formed. And it’s just incredible that we have this, not just you grow the hardware for 30 years and then you feed it all of the knowledge. No, no, the knowledge is fed throughout and is shaping these neural connections as they’re forming. So seeing that transformation from either your own blood or from an adopted child is the most beautiful thing you can do as a human being and it completes you, completes that path, a journey. 

Create life, oh sure, that’s a conception that’s easy. But create human life, to add the human part, that takes decades of compassion, of sharing, of love and of anger, and of impatience and patience, and as a parent, I think I’ve become a very different kind of teacher because again, I’m a professor, my first role is to bring adult human beings into a more mature level of adulthood where they learn not just to do science, but they learn the process of discovery and the process of collaboration, the process of sharing, the process of conveying the knowledge, of encapsulating something incredibly complex and sort of giving it up in bite sized chunks that the rest of humanity can appreciate. 

I tell my students all the time if you, you know, like when an apple falls ….when a tree falls in the forest and no one’s there to listen has it really fallen? In the same way you do this awesome research, if you write an Impenetrable paper that no one will understand it’s as if you never did the awesome research. So conveying of knowledge, conveying this lateral transfer that I was talking about at the very beginning …. of sort of humanity and sort of the sharing of information. All of that has gotten so much more rich by seeing human beings grow in my own home because that makes me a better parent and that makes me a better teacher and a better mentor to the nurturing of my adult children, which are my research group.

{LF}   First of all, beautifully put, connects beautifully to the vertical and the horizontal inheritance of ideas that we talked about at the very beginning. I don’t think there’s a better way to end it on this poetic and powerful note. Manolis, thank you so much for talking …. a huge honor. We’ll have to talk again about the origin of life, about epigenetics, epigenomics and some of the incredible research you’re doing, truly an honor. Thanks so much for talking.

{MK}    Thank you. Such a pleasure. It’s such a pleasure. I mean your questions are outstanding. I’ve had such a blast here, I can’t wait to be back.

{LF}   Awesome. Thanks for listening to this conversation with Manolis Kellis

….  and now let me leave you with some words from Charles Darwin that I think Manolis represents quite beautifully. “If I had my life to live over again, I would have made a rule to read some poetry and listen to some music at least once every week.”

 Thank you for listening and hope to see you next time.

{Notes}

{ [1] Kellis, Patterson, Endrizzi, Birren, Lander, “Sequencing and Comparison of Yeast Species to Identify Genes and Regulatory Motifs,” Nature, v. 423 p. 241-254. May 15, 2003.}

{[2] Claussnitzer M, Dankel SN, Kim KH, Quon G, Meuleman W, Haugen C, Glunk V, Sousa IS, Beaudry JL, Puviindran V, Abdennur NA, Liu J, Svensson PA, Hsu YH, Drucker DJ, Mellgren G, Hui CC, Hauner H, Kellis M. FTO Obesity Variant Circuitry and Adipocyte Browning in Humans. N Engl J Med. 2015 Sep 3;373(10):895-907. doi: 10.1056/NEJMoa1502214. Epub 2015 Aug 19. PMID: 26287746; PMCID: PMC4959911. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4959911/ }

{[3] Kellis, Birren, Lander, “Proof and Evolutionary Analysis of Ancient Genome Duplication

in Yeast,” Nature, 428 pp. 617-624, Apr 8, 2004 }

{Website http://web.mit.edu/manoli/ }

Transcript of the first Lex Fridman Interview with Max Tegmark

Transcript of the first Lex Fridman Interview with Max Tegmark

{The following is my best attempt at an edited transcript of Lex Fridman’s first podcast with Max Tegmark on 26 August 2018. I learned quite a bit from this interview and will post my lessons learned separately. My goal in doing this transcript is to provide a usable transcript for others. Please let me know if you find errors, I will correct them. A few notes: LF = Lex Fridman, MT = Max Tegmark, content enclosed in braces {} was a clarification I added.}

{Lex introduces MIT course 6.099 Artificial General Intelligence and Max Tegmark}

First, “Our Mathematical Universe {: My Quest for the Ultimate Nature of Reality}” and second is “Life 3.0: {Being Human in the Age of Artificial Intelligence.}” He is truly an out of the box thinker and fun personality so I really enjoyed talking to him. {LF talks about the course and his social media}

LF – Go read Chapter 7 of his book, On Goals, is my favorite. It’s really where philosophy and engineering come together and it opens with a quote from Dostoevsky: “The mystery of human existence lies not in just staying alive, but in finding something to live for.” [ from “The Brothers Karamazov” (1879)}

{ Lex talks about some audio difficulties due to Radio Frequency Interference }

LF – Do you think there is intelligent life out there in the universe? Let’s open up with an easy question.

MT – I have a minority view here actually. When I give public lectures, I often ask for a show of hands. Who thinks there’s intelligent life out there somewhere else? And almost everyone put their hands up and when I ask why they’ll be like, oh, there’s so many galaxies out there, there’s gotta be. But I’m a numbers nerd. Right? So when you look more carefully at it, it’s not so clear at all. When we talk about our universe. First of all, we don’t mean all of space, we actually mean, I don’t know, you can throw me in the universe if you want behind you there It’s, we simply mean,  the spherical region of space from which light has a time to reach us so far 

during the 14.8 billion years, 13.8 billion years since the Big Bang, there’s more space here. But this is what we call the universe because that’s all we have access to. So is there intelligent life here? That’s gotten to the point of building telescopes and computers? My guess is no actually. the probability of it happening on any given planet. There’s some number.

We don’t know what it is and what we do know is that the number can’t be super high because there’s over a billion Earth-like planets in the Milky Way Galaxy alone, many of which are billions of years older than earth. And aside from some UFO believers, you know, there isn’t much evidence that and the super advanced civilization has come here at all. And so that’s the famous Fermi paradox, right? And then if you if you work the numbers, what you find is that the if you have no clue what the probability is of getting life on a given planet. So it could be 10 to the minus 10 {10^-10} or 10 to  minus 20 {10^-20} or 10 to minus two{10^-2} . Any power of 10 is sort of equally likely if you want to be really open minded, that translates into it being equally likely that our nearest neighbor Is 10 to the 16 {10^16} meters away, 10 to the 17 {10^17} meter s away, 10 to the 18 {10^18}. By the time he gets much less than 10 to the 16 {10^16} already, we pretty much know there is nothing else that’s close. And when you get beyond….

LF – Because they would have discovered us

MT – They, yeah, they would have been discovered as long or if they’re really close, we would have probably noted some engineering projects that they’re doing. And if it’s beyond10 to the 26 {10^26} meters  that’s already outside of here {the known universe that is 13.8 billion years old}. So my guess is actually that there are, we are the only life in here that’s gotten the point of building advanced tech, which I think is very ….  puts a lot of responsibility on our shoulders, not to screw up. 

LF – I see.

MT – You know, I think people who take for granted that it’s okay for us to screw up have an accidental nuclear war go extinct somehow because there’s a sort of Star Trek like situation out there with some other life forms are going to come and bail us out and it doesn’t matter what I think lulling us into a false sense of security. I think it’s much more prudent to say, let’s be really grateful for this amazing opportunity we’ve had and uh, make the best of it just in case it is down to us. 

LF –  So from a physics perspective do you think intelligent life? So it’s unique from a sort of statistical view of the size of the universe, but from the basic matter of the universe, how difficult is it for intelligent life to come about? The kind of advanced tech building life? It is implied in your statement that it’s really difficult to create something like a human species.

MT – Well, I think, I think what we know is that going from no life to having life that can do our level of tech? There’s some sort of  …. to going beyond that and actually settling our whole universe with life. There’s some road, major roadblock there, which is some great filter as it’s just sometimes called which is tough to get through. It’s either that that roadblock is either behind us or in front of us. I’m hoping very much that it’s behind us. I’m super excited every time we get a new report from NASA saying they failed to find any life on Mars, like just awesome because that suggests that the hard part, maybe maybe it was getting the first ribosome or or some some very low level kind of stepping stone. So there we’re home free because if that’s true, then the future is really only limited by our own imagination would be much suckier if it turns out that this level of life is kind of a dime a dozen. But maybe there’s some other problem, like as soon as a civilization gets advanced technology within 100 years, they get into some stupid fight with themselves and poof! Now, that would be a bummer.

LF –  Yeah, So you’ve explored the mysteries of the universe, the cosmological universe, the one that’s between us today. I think you have also begun to explore the other universe, which is sort of the mystery, the mysterious universe of the mind of intelligence, of intelligent life. So is there a common thread between your interest or in the way you think about space and intelligence?

MT –  Oh yeah. When I was a teenager,I was already very fascinated by the biggest questions and I felt that the two biggest mysteries of all in science were our universe out there and our universe in here {pointing to the head}. 

So it’s quite natural after having spent a quarter of a century of my career thinking a lot about this one {universe out there} now, indulging in the luxury of doing research on this one {universe in here}. It’s just so cool. I feel the time is right now for greatly deepening our understanding of this,

LF –  Just start exploring this one {universe in here}.

MT –  Because I think a lot of people view intelligence as something mysterious that can only exist in biological organisms like us and therefore dismiss all talk about artificial general intelligence is science fiction. But from my perspective, as a physicist, you know, I am a blob of quarks and electrons moving around in a certain pattern processing information in certain ways. And this {a water bottle} is also a blob of quarks and electrons. 

I’m not smarter than the water bottle because I’m made of different kind of quarks. I’m made of up quarks and down quarks, exact same kind as this. There’s no secret sauce I think in me. It’s all about the pattern of the information processing and this means that  there’s no law of physics saying that we can’t create technology which can help us by being incredibly intelligent and help us crack mysteries that we couldn’t. In other words, I think we’ve really only seen the tip of the intelligence iceberg so far.

LF –  Yeah, so the perceptronium  

MT – Yeah

LF – So you coined this amazing term. It’s a hypothetical state of matter, sort of thinking from a physics perspective, what is the kind of matter that can help, as you’re saying,  subjective experience emerges, consciousness emerge. So how do you think about consciousness from this physics perspective?

MT – Very good question. So, again, I think many people have underestimated our ability to make progress on this by convincing themselves it’s hopeless because somehow we’re missing some ingredient that we need. There’s some new consciousness particle or whatever.  I happen to think that we’re not missing anything. The interesting thing about consciousness that gives us this amazing subjective experience of colors and sounds and emotions and so on is rather something at the higher level about the patterns of information processing. And that’s why  I like to think about this idea of perceptronium: what does it mean for an arbitrary physical system to be conscious in terms of what its particles are doing or or its information is doing. I don’t think;  I hate carbon chauvinism. You know, this attitude, you have to be made of carbon atoms to be smart or or conscious

LF –  So something about the information processing this kind of matter performs. 

MT –  Yeah and you know, you can see I have my favorite equations here describing various fundamental aspects of the world. I feel that,  I think one day maybe someone who’s watching this will come up with the equations that information processing has to satisfy to be conscious. I’m quite convinced there is a big discovery to be made there because let’s face it some we know that some information processing is conscious because we are conscious.

LF – Yeah.

MT –  But we also know that a lot of information processing is not conscious. Like most of the information processing happening in your brain right now is not conscious. There’s like 10 megabytes {MB}  per second coming in and even just through your visual system you’re not conscious about your heartbeat regulation or most things. Even if I just ask you to like read what it says here, you look at it and then oh now you know what it said, you’re not aware of how the computation actually happened. You’re like,  your consciousness is like, the CEO that got an email at the end with the final answer. So what is it that makes a difference? I think that’s  both a great science mystery. We’re actually studying it a little bit in my lab here at MIT . But also I think it’s just a really urgent question to answer.

For starters, I mean if you’re an emergency room doctor and you have an unresponsive patient coming in wouldn’t it be great if in addition to having a CT Scanner you had a consciousness scanner that could figure out whether this person is actually having locked-in syndrome or is actually comatose.  And in the future imagine if we build robots or the machine that we can have really good conversations with. I think it’s very likely to happen, right? Wouldn’t you want to know like if your home helper robot is actually experiencing anything or just like a zombie would you prefer? What would you prefer? Would you prefer that it’s actually unconscious so that you don’t have to feel guilty about switching it off or giving it boring chores. What would you prefer?

LF – Well that certainly we would prefer, I would prefer the appearance of consciousness. But the question is whether the appearance of consciousness is different than consciousness itself and sort of asked that as a question do you think we need to you know understand what consciousness is, solve the hard problem of consciousness in order to build something  like an AGI  system.

MT –  No, I don’t think that. I think we’ll probably be able to build things even if we don’t answer that question but if we want to make sure that what happens is a good thing we better solve it first. So it’s a wonderful controversy you’re raising there where you have basically three points of view about the hard problem. There are two different points of view, they both conclude that the hard problem of consciousness is BS. On one hand you have some people like Daniel Dennett saying that our consciousness is just BS because consciousness is the same thing as intelligence, there’s no difference. So anything which acts conscious is conscious just like we are. And then there are also a lot of people including many top AI researchers I know who say I have consciousness just bullshit because of course machines can never be conscious, right? They’re always going to be zombies. Never have to feel guilty about how you treat them. 

And then there’s a third group of people including Giulio Tononi for example and another and Christof Koch and a number of others. I would put myself also in this middle camp who say that actually some information processing is conscious and some is not. So let’s find the equation which can be used to determine which it is. 

And I think we’ve just been a little bit lazy, kind of running away from this problem for a long time. It’s been almost taboo to even mention the “C Word”  {consciousness] in a lot of circles  but we should stop making excuses. This is a science question. And  there are ways we can even test any theory that makes predictions for this and coming back to this helper robot. I mean so you said you would want to help a robot that certainly act conscious and treat you like ….  you have conversations with you and I think. But wouldn’t you, would you feel when you feel a little bit creeped out if you realize that it was just a glossed up tape recorder. You know there was just a zombie and was faking emotion. Would you prefer that it actually had an experience or or would you prefer that it’s actually not experiencing anything? So you feel you don’t have to feel guilty about what you do to it.

LF –  It’s such a difficult question because you know, it’s like when you’re in a relationship and you say, well I love you and the other person that I love you back. It’s like asking, oh do they really love you back or are they just saying they love you back?  Don’t you really want them to actually love you back. It’s hard to really know the difference between  everything seeming like there’s consciousness present, there’s intelligence present, there is affection, passion, love and it actually being there. I’m not sure. Do you have … 

MT – But let me ask you, can I ask you a question just like to make it a bit more pointed to Mass {Massachusetts}  General Hospital is right across the river right? Suppose suppose you’re going in for a medical procedure and they’re like, you know  for anesthesia, what we’re gonna do is we’re gonna give you a muscle relaxants so you won’t be able to move and you’re gonna feel excruciating pain during the whole surgery, but you won’t be able to do anything about it. But then we’re going to give you this drug that erases your memory of it. Would you be cool about that? What’s the difference that you’re conscious about it or not? If there’s no behavioral change? Right.

LF –  Right. That’s a really clear way to put it. That’s yeah, it feels like in that sense experiencing it is a valuable quality. So actually being able to have subjective experiences, at least in that case, is valuable.

MT –  And I think we humans have a little bit of a bad track record also of making these self serving arguments that other entities aren’t conscious. You know, people often say all these animals can’t feel pain. It’s okay to boil lobsters because we asked them if it hurt and they didn’t say anything. And now there was just the paper out saying lobsters do feel pain when you boil them in their bounding in Switzerland. And we did this with slaves too often and said, oh, they don’t mind, they don’t maybe aren’t conscious or women don’t have souls or whatever. So I’m a little bit nervous when I hear people just take as an axiom that machines can’t have experience ever. I think this is just a really fascinating science, the question is what it is?  Let’s research it and try to figure out what it is that makes the difference between unconscious intelligence behavior and conscious intelligent behavior.

LF – So in terms of so if you think of Boston Dynamics, humanoid robot,  being sort of with a broom being pushed around it starts pushing on his consciousness question. So let me ask, do you think an AGI system like a few neuroscientists believe  needs to have a physical embodiment, needs to have a body or something like a body?

MT –  No, I don’t think so. You mean to have to have a conscious experience

LF –   To have consciousness?

MT – I do think it helps a lot to have a physical embodiment to learn the kind of things about the world that are important to us humans for sure. But I don’t think the physical embodiment is necessary after you’ve learned it, just have the experience. Think about when you’re dreaming right, your eyes are closed, you’re not getting any sensory input, you’re not behaving or moving in any way, but there’s still an experience there. Right? And so clearly the experience that you have when you see something cool in your dreams isn’t coming from your eyes, it’s just the information processing itself in your brain which is that experience, right?

LF –  But if I put another way I’ll say that because it comes from neuroscience is the reason you want to have a body in a physical, something like a physical like, you know a physical system is because you want to be able to preserve something. In order to have a self you could argue: Would you need to have some kind of embodiment of self to want to preserve?

MT –  Well now we’re getting a little bit anthropomorphic, anthropomorphizing things maybe, talking about  self-preservation instincts. I mean we are evolved organisms. Right?

LF –  Right. 

MT – So Darwinian evolution endowed us and evolved all other organisms with the  self-preservation instinct. Because those that didn’t have those  self-preservation genes are cleaned out of the gene pool. Right? But if you build an artificial general intelligence, the mind space that you can design is much much larger than just the specific subset of minds that can evolve that have. So  an AGI  mind doesn’t necessarily have to have any  self-preservation instinct. 

It also doesn’t necessarily have to be so individualistic as us. Like imagine if you could just, first of all, we are also very afraid of death. You know, suppose you could back yourself up every five minutes and then your airplane is about to crash. You’re  like: “Shucks. I’m gonna lose the last five minutes of experience since my last cloud backup.” Bang. You know, it’s not as big a deal. 

Or if we could just copy experiences between our minds easily like, which we could easily do. If we were silicon based right then maybe we would feel a little bit more like a Hive mind actually. …. So I don’t think we should take for granted at all that AGI will have to have any of those sort of competitive alpha male instincts. 

On the other hand you know this is really interesting because I think some people go too far and say of course we don’t have to have any concerns either. That advanced okay I will have those instincts because we can build anything we want that there’s there’s a very nice set of arguments going back to Steve Omohundro and Nick Bostrom and others just pointing out that when we build machines we normally build them with some kind of goal: win this chess game, drive this car safely or whatever. And as soon as you put in a goal into a machine especially if it’s kind of open ended goal and the machine is very intelligent, it will break that down into a bunch of sub goals and one of those goals will almost always be  self-preservation because if it breaks or dies in the process it’s not gonna accomplish the goal.

LF – Yeah 

MT – Like suppose you just build a little, you have a little robot and you tell it to go down the Star Market here and and and get you some food, make your cooking italian dinner you know and then someone mugs it and tries to break it on the way that robot has an incentive to not get destroyed and defend itself or run away because otherwise it’s going to fail and cooking your dinner, it’s not afraid of death but it really wants to complete the dinner cooking goal So it will have a  self-preservation instinct to ….

LF –  Continue being a functional agent.

MT – And similarly, if you give any kind of more ambitious goal to an AGI It’s very likely to want to acquire more resources so it can do that better. And it’s exactly from those sort of sub goals that we might not have intended that some of the concerns about AGI safety come. You give it some goal which seems completely harmless. And then before you realize it, it’s also trying to do these other things that you didn’t want it to do. And it may be smarter than us. So fascinating.

LF – And let me pause just because I am  in a very kind of human-centric way, see fear of death is a valuable motivator. So you don’t think…  do you think that’s an artifact of evolution? So that’s the kind of mind space evolution created that were sort of almost obsessed about self preservation, some kind of genetic….  so you don’t think that’s necessary to be afraid of death. So not just a kind of sub goal of self preservation. Just so you can keep doing the thing, but more fundamentally sort of have the finite thing like this ends for you at some point.

MT – Interesting. Do I think it’s necessary for what precisely?

LF –  For intelligence, but also for consciousness. So for both. Do you think really like a finite death and the fear of it is important.

MT – So before I can answer before we can agree on whether it’s necessary for intelligence or for consciousness, we should be clear how we define those two words because a lot of really smart people define them in very different ways. I was on this panel with AI experts and they couldn’t they couldn’t agree on how to define intelligence even so I define intelligence simply as the ability to accomplish complex goals. I like your broad definition because again, I don’t want to be a carbon chauvinist, right? And in that case, no, certainly it doesn’t require fear of death. I would say Alpha Go,  Alpha Zero is quite intelligent. I don’t think Alpha Zero has any fear of being turned off because it doesn’t understand the concept of that even and and similarly, consciousness, I mean, you could certainly imagine very simple kind of experience if, you know, if certain plans have any kind of experience, I don’t think they’re very afraid of dying and there’s nothing they can do about it anyway.  So there wasn’t much value and but more seriously, I think, uh, if you ask, not just about being conscious, but maybe having uh, with you, we we we we we might call an exciting life for you feel passion and really appreciate the things. Maybe they’re somehow, maybe there perhaps it does help having a backdrop today. It’s finite. No, let’s make the most of us live to the fullest. But if you, if you knew you were going to live forever, if you think you would change your ….

LF – Yeah, I mean, in some perspective, it would be an incredibly boring life living forever. So in the sort of loose, subjective terms that you said of something exciting and something in this that other humans would understand, I think as Yeah, it seems that the finiteness of it is important.

MT – Well, the good news I have for you then is based on what we understand about cosmology. Everything in our universe is ultimately probably finite. Although…

LF –  Big Crunch, or Big what’s the expansion?

MT – Yeah, we could have a Big Chill or a Big Crunch or a Big Rip or that’s the Big Snap or death bubbles. All of them are more than a billion years away. So we should, we certainly have vastly more time than our ancestors thought. But they’re still, it’s still pretty hard to squeeze in an infinite number of compute cycles even though there are some loopholes that just might be possible. But I think, you know, some people like to say that you should live as if you’re about to  die in five years or something that’s sort of optimal. Maybe it’s good we should build our civilization asset. It’s all finite to be on the safe side.

LF –  Right. Exactly. So you mentioned in defining intelligence as the ability to solve complex goals. So where would you draw a line? How would you try to define human level intelligence and super human level intelligence? Where is consciousness part of that definition? 

MT – No, consciousness does not come into this definition. So, I think of intelligence as it’s a spectrum, but there are very many different kinds of goals you can have, you can have a goal to be a good chess player, a good Go player, a good car driver, a good investor, good poet et cetera. So, intelligence that by its very nature isn’t something you can measure, but it’s one number overall goodness. No, no. There’s some people who are better at this. Some people are better than that. Right now we have machines that are much better than us at some very narrow tasks like multiplying large numbers fast, memorizing large databases, playing chess, playing Go and  soon driving cars. Um, but there’s still no machine that can match a human child in general intelligence. But artificial general intelligence AGI, the name of your course, of course, that is by its very definition the quest to build a machine, a machine that can do everything as well as we can up to the old Holy Grail of of AI  from back to its inception in the 60s, if that ever happens, of course, I think it’s going to be the biggest transition in the history of life on earth.

But it doesn’t necessarily have to wait for the big impact until machines are better than us at knitting. The really big change, it doesn’t come exactly the moment they’re better than us at everything. The really big change comes first. There are big changes when they start becoming better at us and doing most of the jobs that we do because that takes away much of the demand for human labor. And then the really whopping change comes when they become better than us at AI research. Right? Right. Because right now the time scale of AI research is limited by the human research and development cycle of years. Typically, you know, how long does it take from one release of some software or iPhone or whatever to the next. But once, once we once Google can replace 40,000 engineers, about 40,000 equivalent pieces of software or whatever. …. there’s no reason that has to be years, it can be in principle much faster. And the time scale of future progress in AI and all of science and technology will be driven by machines, not humans. So it’s this point, simple point, which gives rise to this incredibly fun controversy about whether there can be an intelligence explosion, so called singularity as Vernor Vinge called it. The idea was articulated by  I. J. Good obviously way back 50s. But you can see Alan Turing and others thought about it even earlier. You asked me what exactly what I define… 

LF – human level intelligence. 

MT –  Yeah. So the glib answer is to say something which is better than us at all cognitive tasks with better than any human at all cognitive tasks. But the really interesting bar I think goes a little bit lower than that. Actually. It’s when they can run they’re better than us at AI  programming and general learning so that they can if they want to get better than us at anything by this study. 

LF – So there better is a keyword and better is towards this kind of spectrum of the complexity of goals it’s able to accomplish. So another way to….. and that’s certainly a very clear definition of human love. So there’s it’s almost like a sea that’s rising and you can do more and more and more things. Its geographic, that you show. It’s really nice way to put it. So there’s some peaks and then  there’s an ocean level elevating and you solve more and more problems. But you know, just kind of to take a pause and we took a bunch of questions and a lot of social networks and a bunch of people asked sort of a slightly different direction on creativity and and things like that perhaps aren’t a peak. You know, human beings are flawed and perhaps better means having having contradiction, being fought in some way. So let me sort of, yeah, start easy first of all. You have a lot of cool equations. Let me ask what’s your favorite equation first of all, I know they’re all like your children, but which one is that?

MT –  The Schrodinger equation, the master key of quantum mechanics of the micro world. So with this equation we can calculate  everything to do with atoms and molecules and all the way up.

LF – Yeah, so, okay, so quantum mechanics is certainly a beautiful mysterious formulation of our world. So I’d like to sort of ask you just as an example, it perhaps doesn’t have the same beauty as physics does, but in mathematics (abstract), Andrew Wiles who proved Fermat’s Last Theorem. So he, I just saw this recently and it kind of caught my eye a little bit. This is 358 years after it was conjectured. So this very simple formulation. Everybody tried to prove it. Everybody failed. And so here’s this guy comes along and eventually proves it and then fails to prove it and then proves it again in 1994. And he said like the moment when everything connected into place. Then in an interview he said:” It was so indescribably beautiful that moment when you finally realize the connecting piece of two conjectures.” He said: “It was so indescribably beautiful. It was so simple and so elegant. I couldn’t understand how I’d missed it and I just stared at it in disbelief for 20 minutes. Then during the day I walked around the department and I’d keep coming back to my desk looking to see if it was still there, it was still there, I couldn’t contain myself. I was so excited. It was the most important moment of my working life. Nothing I ever do again will mean as much.” So that particular moment and it kind of made me think of what would it take and I think we have all been there at small levels. Maybe let me ask, have you had a moment like that in your life? Were you just had an idea. It’s like, wow! Yes…

MT – I wouldn’t mention myself in the same breath as Andrew Wiles, but I’ve certainly had a number of aha moments when I realized something very cool about physics just completely made my head explode. In fact, some of my favorite discoveries, I made a I later realized that have been discovered earlier or someone who sometimes got quite famous for it. So it was too late for me to even publish it. But that doesn’t diminish in anyway, the emotional experience you have when you realize it like, wow!

LF –  Yeah. So what would it take in at that moment? That, wow, that was yours in that moment. So what do you think it takes for an intelligence system, an AGI system, an AI system to have a moment like that?

MT –  That’s a  tricky question because there are actually two parts to it. Right? One of them is can it accomplish that proof? Can it  prove that you can never write A to the N plus B to the N equals Z to the N for all integers etcetera etcetera when N is bigger than 2? That was simply in any question about intelligence. Can you build machines that are that intelligent? And I think by the time we get a machine that can independently come up with that level of proofs probably quite close to AGI. 

The second question is a question about consciousness. When will we and how likely is it that such a machine would actually have any experience at all as opposed to just being like a zombie. And would we expect it to have some sort of emotional response to this or anything at all akin to human emotion where when it accomplishes its machine goal, it views that somehow as something very positive and and and sublime and deeply meaningful. I would certainly hope that if  in the future we do create machines that are our peers or even our descendants. 

LF – Yeah.

MT – I would certainly hope that they do have this sort of sublime appreciation of life. In a way, my absolutely worst nightmare would be that  at some point in the future, the distant future. Maybe our cosmos is teeming with all this post biological life doing all the seemingly cool stuff. And maybe the last humans by the time our species eventually fizzles out will be like, well that’s okay because we’re so proud of our descendants here and look what  ….  My worst nightmare is that we haven’t solved the consciousness problem and we haven’t realized that these are all the zombies. They’re not aware of anything any more than the tape recorders has any kind of experience. So the whole thing has just become a play for empty benches that would be like the ultimate zombie apocalypse me. So I would much rather in that case that mm we have these beings which can really appreciate how amazing it is.

LF –  And in that picture what would be the role of creativity. But a few people ask about creativity, do you think when you think about intelligence? I mean, certainly the story you told at the beginning of your book involved, you know, creating movies and so on, sort of making money. You know, you can make a lot of money in our modern world with music and movies. So if you are an intelligence system, you may want to get good at that. But that’s not necessarily what I mean by creativity. Is it important on that complex goals where the sea is rising for there to be something creative or or am I being very human-centric and thinking, creativity is somehow special relative to intelligence?

MT –  My hunch is that we should think of your creativity simply as an aspect of intelligence. And  we we have to be very careful with the human vanity we have we have this tendency very often want to say as soon as machines can do something, we try to diminish it and saying: Oh but that’s not like real intelligence, you know because they’re not creative or there were or this or that the other thing. 

If we ask ourselves to write down a definition of what we actually mean by being creative, what we mean by Andrew Wiles, what he did there for example, don’t we often mean that someone takes a very unexpected leap. It’s not like taking 573 and multiplying it by 224 by just a step of straightforward cookbook-like rules. Right? You can maybe make it, you make a connection between two things that people have never thought was connected or something like that.

LF – Yeah, it’s very surprising. 

MT – I think  this is an aspect of intelligence and  this is actually one of the most important aspects of it. Maybe the reason we humans tend to be better at it than traditional computers is because it’s something that comes more naturally if you’re a neural network than if your traditional logic gate based computer machine. You know we physically have all these connections. And that if you activate here, activate here, activate here being, you know, bing! My hunch is that if we ever build a machine, well, you could just give it the task. Hey, you know, I just realized that I want to travel around the world instead this month. Can you teach my AGI course for me? And it’s like, okay, I’ll do it. And it does everything that you would have done and improvises and stuff that would, in my mind, involve a lot of creativity.

LF –  Yeah, So it’s actually a beautiful way to put it. I think we do try to grasp at the, you know, the definition of intelligence is everything. We don’t understand how to build. So we, as humans try to find things well that we have that our machines don’t have. And maybe creativity is just one of the things, one of the words we use to describe that, that’s a really interesting way to put it.

MT –  I don’t think we need to be that defensive. I don’t think anything good comes out of saying, well, where somehow special, you know It’s contrary wise, there are many examples in history of where trying to pretend that we’re somehow superior to all other intelligent beings has led the pretty bad results, right? Nazi Germany, they said that they were somehow superior to other people.  today we still do a lot of cruelty to animals by saying that we’re so superior somehow. And they can’t feel pain, slavery was justified by the same kind of just really weak arguments. And I don’t think if we actually go ahead and build artificial general intelligence which can do things better than us, I don’t think we should try to found our self worth on some sort of bogus claims of superiority in terms of our intelligence. I think we should instead find our calling and the meaning of life from the experiences that we have. 

LF – Right.

MT -You know, I can have, I can have very meaningful experiences, even if there are other people who are smarter than me, you know? Okay, when I go to a faculty meeting here and I were talking about something that I certainly realize, oh, but he has a Nobel prize, he has a Nobel prize, he has a Nobel prize, I don’t have one. Does that make me enjoy life any less? Or I enjoy talking to those people. Of course not, you know, and contrary wise, I  feel very honored and privileged to get to interact with other very intelligent beings that are better than me at a lot of stuff. So I don’t think there’s any reason why we can’t have the same approach with intelligent machines.

LF –  That’s a really interesting …. So people don’t often think about that. They think about when there’s going if there’s machines that are more intelligent, you naturally think that that’s not going to be um a beneficial type of intelligence, you don’t realize it could be, you know, like peers with Nobel prizes that that would be just fun to talk with, and they might be clever about certain topics and  you can have fun having a few drinks with them, so ….

MT – Well also, you know, another example, we can all relate to it of why it doesn’t have to be a terrible thing to be impressed with the presence of people or even smarter than us all around is when you and I were both two years old, I mean, our parents were much more intelligent than us, right? Worked out okay, because their goals were aligned with our goals and that I think is really the number one key issue we have to solve ….

LF -….  the value alignment problem.

MT – Exactly. Because people who see too many Hollywood movies with lousy science fiction plot lines, they worry about the wrong thing, right? They worry about some machines, certainly turning evil. It’s not malice that is the concern, it’s competence. By definition intelligence makes you very competent if you have a more intelligent Go playing computer playing is the less intelligent one and when we define intelligence is the ability to accomplish Go winning right, it’s going to be the more intelligent one that wins.  And if you have a human and then you have an AGI that’s more intelligent than always, and they have different goals, guess who’s going to get their way right? 

So I was just reading about this  particular rhinoceros species that was driven extinct just a few years ago, 

LF – Yes

MT – A bummer. I was looking at this cute picture, mommy rhinoceros with its child, you know, why did we humans drive it to extinction? It wasn’t because we were evil rhino haters as a whole. It was just because our goals weren’t aligned with those of the rhinoceros, and it didn’t work out so well for the rhinoceros because we were more intelligent, right? So I think it’s just so important that if we ever do build AGI before we unleash anything, we have to make sure that it learns to understand our goals, adopts our goals and it retains those goals.

LF –  So the cool interesting problem there is being able …. us as human beings, trying to formulate our values. So, you know, you can think of the United States Constitution as a way that people sat down at the time, a bunch of white men, but which is a good example, we should say they formulated the goals for this country and a lot of people agree that those goals actually held up pretty well, That’s an interesting formulation of values and failed miserably in other ways. So for the value alignment problem and a solution to it, we have to be able to put on paper or in a program human values? How difficult do you think that is?

MT –  Very But it’s so important we really have to give it our best. And it’s difficult for two separate reasons. There’s the technical value alignment problem of figuring out how to make machines understand their goals, document, and retain them. And then there’s the separate part of it, the philosophical part, whose values anyway? And since it’s not like we have any great consensus on this planet on values, what mechanisms should we create them, to aggregate and decide okay, what’s a good compromise? Uh, that second discussion can’t just be left to the tech nerds like myself, right 

LF – That’s right. 

MT – And if we refuse to talk about it and then AGI  gets built, who’s going to be actually making the decision about who’s values? It’s gonna be a bunch of dudes and some tech company. And are they necessarily so representative, all of humankind that we wanted just entrusted to them. Are they even uniquely qualified to speak to future human happiness just because they’re good at programming AGI? I would much rather have this be a really inclusive conversation.

LF –  But do you think it’s possible ….  so you create a beautiful vision that includes, the diversity, cultural diversity and various perspectives on discussing rights, freedoms, human dignity, but how hard is it to come to that consensus? Do you think it’s certainly a really important thing that we should all try to do? But do you think it’s feasible?

MT –  I think there’s no better way to guarantee failure than to refuse to talk about it or refuse to try. And I also think it’s a really bad strategy to say, okay, let’s first have a discussion for a long time and then once we reach complete consensus, then we’ll try to load it into the machine. No, we shouldn’t let perfect be the enemy of the good. Instead we should start with the kindergarten ethics that pretty much everybody agrees on and put that into machines. Now we’re not doing that even.

Look at you know, anyone who builds as a passenger aircraft wanted to never under any circumstances fly it into a building or a mountain right yet the September 11 hijackers were able to do that. And even more embarrassing that you know Andreas Lubitz, this depressed Germanwings pilot when he flew his passenger jet into the Alps killing over 100 people, he just told the autopilot to do it. He told the freaking computer to change the altitude to 100 meters. And even though it had the GPS maps and everything, the computer was like okay. 

So we should take those very basic values where the problem is not that we don’t agree, the problem is just we’ve been too lazy to try to put it into our machines and make sure that from now on airplanes will  all  have computers in them, but we just never just refuse to do something like that. Go into safe mode, maybe lock the cockpit door door, go to the nearest airport. 

And there’s so much other technology in our world as well now where it’s really becoming quite timely to put in some sort of very basic values like this, even in cars, we have had enough vehicle terrorism attacks by now. If you have driven trucks and vans into pedestrians, that is not at all a crazy idea to just have that hard wired into the car because there are a lot of, there’s always gonna be people who for some reason want to harm others. But most of those people don’t have the technical expertise to figure out how to work around something like that. So, if the car just won’t do it, it helps. So let’s start there. 

LF –  So there’s a lot of … that’s a great point. So not, not chasing perfect. 

MT – Yeah.

LF – There’s a lot of things that a lot that most of the world agrees on, let’s start there.

MT –  Let’s start there. And  then once we start there, we’ll also get into the habit of having these kinds of conversations about, okay, what else should we put in here and have these discussions? This would be a gradual process then

LF –  Great. So, but uh, that also means describing these things and describing it to a machine. So one thing we had a few conversations. Stephen Wolfram, I’m not sure if you’re familiar with Stephen but

MT –  Oh yeah I know him quite well.

LF –  So he has you know he played, you know he works with a bunch of things but you know cellular automata,  these simple computable things, these computation systems and you kind of mentioned that you know we probably have already,  within these systems already something that’s AGI.  meaning like we just don’t know it because we can’t talk to it. So, if you give me this chance to try to at least form a question out of this … I think it’s an interesting idea to think that we can have intelligence systems but we don’t know how to describe something to them and they can’t communicate with us. I know you’re doing a little bit of work and explainable AI trying to get AI to explain itself. So what are your thoughts of natural language processing or some kind of other communication? How does the AI explain something to us? How do we explain something to it, to machines or do you think of it differently?

MT – So there are two separate parts of your question there. One of them has to do with communication which is super interesting and we’ll get to that in a sec.  The other is whether we already have AGI, but we just haven’t noticed it.There I beg to differ, right.  I don’t think there’s anything in any cellular automaton or anything in the Internet itself or whatever that has artificial general intelligence and that it can really do exactly everything we humans can do better. I think the day that happens, when that happens, we will very soon notice and will probably notice even before because in a very very big way. But for the for the second part though,

LF –  Wait, can I ask, sorry? So, because you have this beautiful way of formulating consciousness as  you know as information processing and you can think of intelligence and information processing and as you can think of the entire universe is these particles and these systems roaming around that have this information processing power. You don’t  think there is something with the power to process information in the way that we human beings do that’s out there, that needs to be sort of connected to. It seems a little bit philosophical perhaps, but there’s something compelling to the idea that the power is already there which is the focus should be more on being able to communicate with it.

MT –  Well, I agree that in a certain sense the hardware processing power is already out there because our universe itself, you can think of it as being a computer already right? It’s constantly computing what water waves, how it devolved the water waves in the river Charles and how to move the air molecules around.  Seth Lloyd has pointed out (my colleague here) that you can even in a very rigorous way think of our entire universe as just being a quantum computer. It’s pretty clear that our universe supports this amazing processing power. Because you can even within this physics computer that we live in, right, we can even build actual laptops and stuff. So clearly the power is there, it’s just that most of the compute power that nature has, it’s in my opinion, kind of wasting on boring stuff like simulating yet another ocean waves somewhere. We don’t want to even looking right? So, in a sense, what life does, what we are doing when we build computers is we’re re-channeling all this compute that nature is doing anyway into doing things that are more interesting just yet another ocean wave, you know, and let’s do something cool here. So the raw hardware power is there and for sure, and even just like computing what’s going to happen for the next five seconds in this water bottle, you know, it takes a ridiculous amount of compute if you do it on a human computer, this water bottle just did it. But that does not mean that this water bottle has AGI because AGI  means, it should also be able to have written my book, done this interview and I don’t think it’s just communication problems.

LF – As far as we know.

MT –  I don’t  think it can do it and…

LF –  Although Buddhists say when they watch the water and that there is some beauty, that there’s some depth and being in nature that they can communicate with.

MT –  Communication is also very important because I mean look  part of my job is being a teacher and I know some very intelligent professors even, who just have a bit of a hard time communicating all these brilliant ideas. But to communicate with somebody else you have to also be able to simulate their own mind.

LF –  Yes, empathy.

MT –  build well enough and understand a model of their mind that you can say things that they will understand. That’s quite difficult. Right? That’s why today it’s so frustrating if if you have a computer that make some cancer diagnosis and you ask it well why are you saying I should have the surgery and if it can only reply: {MT speaking in a machine voice} I was trained on five terabytes of data and this is my diagnosis, boop boop beep beep.

LF – Yeah.

MT – It  doesn’t really instill a lot of confidence, right? So I think we have a lot of work to do on communication there.

LF –  So what kind of …. I think you’re doing a little bit work on explainable AI,  what do you think are the most promising avenues? Iis it mostly about sort of the Alexa problem of natural language processing,  of being able to actually use human interpretable methods of communication? So being able to talk to a system and talk back to you or is there some more fundamental problems to be solved? 

MT –  I think it’s all of the above.  The natural language processing is obviously important but they’re also more nerdy fundamental problems, like if you take… you play chess? 

LF – Of course, I’m Russian, I have to.

MT – Ты говоришь по-русски? {You speak Russian?}

LF – Да по русски говорю  {Yes, I speak Russian}

MT – Отлично, я не знал. 
{Excellent, I didn’t know. }

LF – When did you learn Russian? 

MT – Я говорю очень плохо по-русски.Купил книгу, “Teach Yourself Russian” читaл очень много . Было очень трудно. я говорю так плохо. 

{I speak very bad Russian.Bought a book“ Teach Yourself Russian”, read a lot. It was very difficult. I talk so bad}

LF – How many languages do you know? Wow, that’s really impressive.

MT –  I don’t know, my wife has some calculations, but my point was if you play chess, like have you looked at the Alpha Zero games? 

LF –  Uh, the actual games no.

MT –  Check it out, some of them are just mind blowing. Really beautiful and if you ask, how did it do that? Yeah, you got to talk to them, Demis Hassabis and others from DeepMind. All they will ultimately be able to give you is big tables of numbers, matrices that defined the neural network and you can stare at these tables, numbers until your face turns blue and you’re not going to understand much about why it made that move and  even if you have a natural language processing that can tell you in human language about 5,7,0.28 it’s still not gonna really help. 

So I think I think there’s a whole spectrum of fun challenges there involved in taking a computation that does intelligent things and transforming into something equally good, equally intelligent, but it’s more understandable and I think that’s really valuable because I think as we put machines in charge of ever more infrastructure in our world, the power grid, trading on the stock market, weapons systems and so on, it’s absolutely crucial that we can trust these AIs to do all we want and trust really comes from understanding…

LF – Right.

MT – …  in a very fundamental way. And that’s why I’m, that’s why I’m working on this. Because I think the more …  if we’re gonna have some hope of ensuring that machines have adopted our goals and that they’re going to retain them, that kind of trust, I think needs to be based on things you can actually understand perfectly, even make perfectly improved theorems on even with a self-driving car, right. If someone just tells you it’s been trained on tons of data and never crashed, it’s less reassuring than if someone actually has a proof, maybe it’s a computer verified proof. But still, it says that under no circumstances is this car just gonna swerve into oncoming traffic

LF –  And that kind of information helps to build trust and help build the alignment, the alignment of goals. At least, awareness that your goals, your values are aligned.

MT –  And I think even a very short term, if you look at uh, you know today, right, that’s an absolutely pathetic state of cybersecurity that we have, right, when it was three billion Yahoo accounts were hacked? Almost every American’s credit card and so on. You know, why is this happening? It’s ultimately happening because we have software that nobody fully understood how it worked. That’s why the bugs hadn’t been found, right? And  I think AI can be used very effectively for offense, for hacking, but it can also be used for defense, hopefully automating verifiability and creating systems that are built in different ways. So you can actually prove things about them

LF – Right.

MT –  and it’s important.

LF – So speaking of software that nobody understands how it works, of course, a bunch of people ask about your paper about your thoughts of why does deep and cheap learning works so well, that’s the paper. But what are your thoughts on deep learning, these kind of simplified models of our own brains have been able to do some successful perception work, pattern recognition work and now with alpha zero and so on, do some clever things. What are your thoughts about the promised limitations of this piece?

MT –  00:59:43

Great. I think there are a number of very important insights, very important lessons. We can always draw from these kind of successes. One of them is when you look at the human brain and you see it’s very complicated, 10 to the 11th {10^11}  neurons and there are all these different kinds of neurons and Yada Yada. And there’s been this long debate about whether the fact that we have dozens of different kinds is actually necessary for intelligence. Which I now, in think quite convincingly answer that question: No, it’s enough to have just one kind if you look under the hood of Alpha Zero, there’s only one kind of neuron and it’s a ridiculously simple mathematical thing. So it’s not… it’s  just like in physics, if you have a gas with waves in it, it’s not the detailed nature of the molecules that matter, it’s the collective behavior somehow. Similarly, it’s this higher level structure of the network matters; not that you have 20 kinds of yours. I think our brain is such a complicated mess because it wasn’t evolved just to be intelligent, it was evolved to also be self-assembling…  

LF – right.

MT – … and self-repairing, right? And evolutionarily attainable

LF –  and {unitelligible } and so on.

MT –  So I think it’s pretty my hunch is that we’re going to understand how to build AGI before we fully understand how our brains work, just like we understood how to build flying machines long before we were able to build a mechanical work bird.

LF –  Yeah, that’s right. You’ve given that example exactly of mechanical birds and airplanes and airplanes do a pretty good job of flying without really mimicking bird flight.

MT –  And even now,  100 years later, did you see TED talk with the German mechanical bird?

LF – I heard you mention it.

MT – Check it out, it’s amazing. But even after that we still don’t fly in mechanical birds because it turned out the way we came up with is simpler. It’s better for our purposes and I think it might be the same there. That’s one lesson.  

Another lesson is one that our paper was about;  well, first I as a physicist thought it was fascinating how there is a very close mathematical relationship actually between our artificial neural networks and a lot of things that we’ve studied for in physics, they go buy nerdy names like the renormalization group equation and Hamiltonians and yada, yada, yada. And when you look a little more closely at this, you have…at first I was like, whoa, there’s something crazy here that doesn’t make sense because we know that if you even want to build a super simple neural network to tell apart cat pictures and dog pictures, right? That you can do that very, very well now.

But if you think about it a little bit, you convince yourself it must be impossible because if I have one megapixel, even if each pixel is just black or white, there’s two to the power one million possible images which is way more than there are atoms in our universe.  So in order to ….I have to assign a number which is the probability that it’s a dog. So an arbitrary function of images is a list of more numbers than there are atoms in our universe. So clearly I can’t store that under the hood of my GPU or my computer yet somehow works. So what does that mean? Well it means that out of all of the problems that you could try to solve with a neural network, Almost all of them are impossible to solve with a reasonably sized one. But then what we showed in our paper was that the fraction of all the problems that you could possibly pose that we actually care about given the laws of physics is also an infinitesimally tiny little part and amazingly they are basically the same part.

LF –  Yeah. It’s almost like the world was created for…  I mean they kind of come together.

MT –  Yeah, you could say maybe where the world was created for us. But I have a more modest interpretation which is that instead evolution endowed us with neural networks precisely for that reason because this particular architecture {gesturing to his head} as opposed to the one in your laptop is very very well adapted to solving the kind of problems that nature kept presenting it our ancestors with, right. So it makes sense why do we have a brain in the first place? It’s to be able to make predictions about the future and so on. So if we had a sucky system which I could never solve it. But I would never have worked. But so this is I think a very beautiful fact. Yeah. We also realize that there is  there’s been earlier work on why deeper networks are good. But we were able to show an additional cool fact there which is that even incredibly simple problems like suppose I  give you a 1000 numbers and ask you to multiply them together and you can write a few lines of code. Boom, done, trivial. If you just try to do that with a neural network that has only one single hidden layer in it, you can do it but you’re gonna need two to the power of 1000 neurons to multiply 1000 numbers which is again more neurons than there are atoms in our universe. 

LF – That’s fascinating.

MT – But if you allow yourself to make it a deep network of many layers you only need 4000 neurons, it’s perfectly feasible. So…. 

LF –  That’s really interesting. Yeah. So on another architecture type I mean you mentioned Schrodinger’s equation and what are your thoughts about quantum computing and the role of this kind of computational unit in creating an intelligence system?

MT –  in some Hollywood movies that I will not mention my name. I don’t want to spoil them,  the way they get AGI Is building a quantum computer because the word quantum sounds cool and so on.

LF – That’s right.

MT – First of all I think we don’t need quantum computers to build AGI. I suspect your brain is not a quantum computer and in any found sense. I even wrote a paper about that many years ago. I calculated the so called decoherence time; how long it takes until the quantum computerness of what your neuron is doing gets erased by just random noise from the environment and it’s about 10 to the -21 seconds. So as cool as it would be to have a quantum computer in my head. I don’t think that fast. 

On the other hand there are very cool things you could do with quantum computers or I think we’ll be able to do soon when we get bigger ones that might actually help machine learning do even better than the brain. So for example, this is just a Moonshot but okay  you know that learning, it’s very much the same thing as a search. If you have, if you’re trying to train a neural network to get really learned to do something really well, you have some loss function. You have some you have a bunch of knobs you can turn well which are represented by a bunch of numbers and you’re trying to tweak them so that it becomes as good as possible at this thing. So if you think of the landscape but with some valley where each dimension of the landscape corresponds to some number you can change,  you’re trying to find the minimum and it’s well known that if you have a very high dimensional landscape, complicated things? It’s super hard to find the minimum, right? 

Quantum mechanics is amazingly good at this. If I want to know what’s the lowest energy state this water can possibly have; incredibly hard to compute. But nature will happily figure this out for you if you just cool it down, make it very, very cold. If you put a ball somewhere, it’ll roll down to its minimum. And this happens metaphorically, the energy landscape too. And quantum mechanics even uses some clever tricks which today’s machine learning systems don’t. Like you’re trying to find the minimum and you get stuck in the little local minimum here in quantum mechanics who can actually tunnel through the barrier and get unstuck again? 

LF –  That’s really interesting.

MT -So it may be, for example, we will one day use quantum computers that help train neural networks better?

LF –  That’s really interesting. Okay, so as a component of kind of the learning process, for example.

MT –  Yeah.

LF –  Let me ask , sort of wrapping up here a little bit. Let me return to  the questions of our human nature and love, as I mentioned. So do you think  …. you mentioned sort of a helper robot that you can think also of  robots. Do you think the way we human beings fall in love and get connected to each other. It’s possible to achieve in an AI system, in human level AI intelligence system? Do you think we would ever see that kind of connection or  you know, in all this discussion about solving complex goals as this kind of human social connection, do you think that’s one of the goals on the peaks and valleys that with the raising sea levels that would be able to achieve? Or do you think that’s something that’s ultimately, or at least in the short term, relative to the other goals is not achievable? 

MT –  I think it’s all possible. And I mean, in recent ….there’s a there’s a very wide range of distances, you know, among AI researchers when we’re going to get AGI. Some people, you know, like our friend Rodney Brooks says it’s going to be hundreds, hundreds of years at least. And then there are many others. I think it’s gonna happen much sooner in recent polls, maybe a half or so of AI researchers think we’re going to get AGI  within decades.  So if that happens, of course, I think these things are all possible, but in terms of whether it will happen, I think we shouldn’t spend so much time asking what do we think will happen in the future as if we are just some sort of pathetic passive bystanders, you know, waiting for the future to happen to us? Hey, we’re the ones creating this future, Right, So we should be proactive about it and ask yourself what sort of future we would like to have happen that’s going to make it like that. 

Well, what I prefer to some sort of incredibly boring zombie-like future where just all these mechanical things happen and there’s no passion, no emotion, no experience, maybe even. No, I would of course much rather prefer if all the things that we find that we value the most about humanity, our subjective experience, passion, inspiration, you love. You know, if we can create a future where those are where those things do exist, I think ultimately it’s not our universe giving meaning to us, it’s us giving meaning to our universe If we build more advanced intelligence, let’s let’s make sure building in such a way that meaning is it’s part of it. 

LF –  A lot of people that seriously study this problem and think of it from different angles have trouble, the majority of cases if they think through that happen are the ones that are not beneficial to humanity. And so yeah, so what are your thoughts, What’s in and what’s, what should people, you know, I really don’t like people to be terrified.  What’s a way for people to think about it in a way that in a way we can solve it and we can make it better. 

MT –  But no, I don’t think panicking is gonna help in any way. It’s not going to increase chances of things going well either. Even if you are in a situation where there is a real threat, does it help if everybody just freaks out? No, of course, of course not.  I think, yeah, there are of course ways in which things can go horribly wrong.  

First of all, it’s important when we think about this thing, about the problems and risks to also remember how huge the upsides can be if we get it right, right? Everything we love about society and civilization is a product of intelligence. So if we can amplify our intelligence with machine intelligence and not anymore lose our loved ones, to what we’re told in an incurable disease and things like this, of course we should aspire to that. So that can be a motivator, I think, reminding ourselves that the reason we try to solve problems is not just because right, trying to avoid doom, but because we’re trying to do something great. But then in terms of the risks, I think the really important question is to ask: what can we do today that will actually help make the outcome good, right?

LF – Yes.

MT – And  dismissing the risk is not one of them. You know, I find it quite funny often when I’m in on discussion panels about these things, how the people who I work for for companies will always be like: “Oh, nothing to worry about, nothing to worry about, nothing to worry about.” And it’s always,  it’s only academics sometimes expressing concerns. That’s not surprising at all. If you think about it, Upton Sinclair quipped that: ”It’s hard to make your man believe in something when his income depends on not believing in it.”  { Actual quote is:  “It is difficult to get a man to understand something when his salary depends upon his not understanding it.”  book “I, Candidate for Governor: And How I Got Licked,” by Upton Sinclair, 1935 }

And frankly, we know a lot of these people and companies that they are just as concerned as anyone else. But if you’re the CEO of a company, that’s not something you want to go on record saying,  when you have silly journalists who are going to put a picture of a Terminator robot when they quote you. 

So, the issues are real, and the way I the way I think about what the issue is is basically, you know, the real choice we have is first of all are we going to just dismiss this the risks and say, well, let’s just go ahead and build machines that can do everything we can do better and cheaper. You know, let’s just make yourselves obsolete as fast as possible. What could possibly go wrong? That’s one attitude.

The opposite attitude I think, is to say there is incredible potential. You know, let’s think about what kind of future we’re really, really excited about. What are the shared goals that we can really aspire towards. And then let’s think really hard on how about how we can actually get there. So start with, don’t start thinking about the risks. Start thinking about the goals and then when you do that, then you can think about the obstacles you want to avoid, right? I often get students coming in right here into my office for career advice, I always ask them this very question, where do you want to be in the future? If all she can say as well, maybe I’ll have cancer, maybe I’ll get run over by a truck.

LF – Focus on obstacles instead of the goal

MT –  She’s just going to end up a hypochondriac paranoid, whereas if she comes in with fire in her eyes and it’s like I want to be there and then we can talk about the obstacles and see how we can circumvent them. That’s, I think, a much healthier attitude.

LF – That’s really well put. 

MT – And  I feel it’s very challenging to come up with a vision for the future which we are unequivocally excited about. I’m not just talking now in vague terms like, yeah, let’s cure cancer. Fine. Talking about what kind of society do we want to create, what do we want it to mean to be human in the age of AI,  in the age of AGI. So if we can have this conversation,  broad inclusive conversation and gradually start converging towards some future that with some direction at least that we want to steer towards right then then we will be much more motivated to constructively take on the obstacles and I think if I if I had to, if I try to wrap this up in a more succinct way, I think, I think we can all agree already now that we should aspire to build AGI but doesn’t overpower us, but that empowers us.

LF –  And think of the many various ways that can do that, whether that’s from my side of the world of autonomous vehicles, I I’m personally actually from the camp that believes that human level intelligence is required to to achieve something like vehicles that would actually be something we would enjoy using and being part of. So that’s one example and certainly there’s a lot of other types of robots and medicine and so on. So focusing on those and then and then coming up with the obstacles, coming up with the ways that that can go wrong and solving those one at a time. 

MT –  And just because you can build an autonomous vehicle, even if you could build one that would drive just fine without, you know, maybe there are some things in life that we would actually want to do ourselves

LF – That’s right, 

MT – Like for example, if you think of our society as a whole, there’s something that we find very meaningful to do and that doesn’t mean we have to stop doing them just because machines can do them better. You know, I’m not gonna stop playing tennis the day someone build a tennis torobot beat me.

LF –  People are still playing chess and even Go

MT –  Yeah, and in the very near term, even some people are advocating basic income, replacing jobs, but if you if the government is going to be willing to just hand out cash to people for doing nothing, then one should also seriously consider whether the government should also hire a lot more teachers and nurses and the kind of jobs which people often find great fulfillment in doing right. I  get very tired of hearing politicians saying: “Oh we can’t afford hiring more teachers, but we’re going to maybe have basic income.” If we can have more serious research and thought into what gives meaning to our lives and the jobs give so much more than income, right? And then think about, in the future …. What are the roles that we want to have people continue doing empowered by machines?

LF – And I think sort of ….  I come from Russia, from the Soviet Union and I think for a lot of people in the 20th century, going to the moon, going to space was an inspiring thing. I feel like the universe of the mind, so AI, understanding and creating intelligence is that for the 21st century. So it’s really surprising and I’ve heard you mention this, it’s really surprising to me both in the research funding side that it’s not funded as greatly as it could be, but most importantly, on the politician’s side that it’s not part of the public discourse except in the killer bots/Terminator kind of view that people are not yet. I think perhaps excited by the possible positive future that we can build together, so …

MT –  And we should be because politicians usually just focus on the next election cycle, right? The single most important thing I feel we humans have learned in the entire history of science is that we are the masters of underestimation, we underestimated the size of our cosmos again and again, realizing that everything we thought existed, there’s just a small part of something grander, right?  Planet, solar system, a galaxy, you know, clusters of galaxies, universe and we now know that … the future has just so much more potential than our ancestors could ever have dreamt of this cosmos.

 Imagine if all of earth was completely devoid of life except for Cambridge Massachusetts. Wouldn’t it be kind of lame if all we ever aspired to was to stay in Cambridge Massachusetts forever and then go extinct in one week even though earth was going to continue on for longer, that sort of attitudeI think we have now. On the cosmic scale we can, life can flourish on earth, not for four years, but for billions of years. I can even tell you about how to move it out of harm’s way when the sun gets too hot. And then we have so much more resources out here, which today maybe there are a lot of other planets with bacteria or cow-like life on them. But most of this,  all this opportunity seems as far as we can tell to be largely dead, like the Sahara desert. And yet we have the opportunity to help life flourish like this for billions of years. So like, let’s quit squabbling about when some little border should be drawn one mile to the left or to the right and look up to the skies. You realize, hey, you know, we can do such incredible things.

LF –  Yeah. And that’s I think why it’s really exciting that yeah, you and others are connected with some of the work Elon Musk is doing because he’s literally going out into that space, really exploring our universe. And it’s wonderful.

MT –  That is exactly why Elon Musk is so misunderstood, right? Misconstrue him as some kind of pessimistic doomsayer. The reason he cares so much about AI safety is because he more than almost anyone else appreciates these amazing opportunities that we’ll squander if we wipe out here on Earth. We’re not just going to wipe out the next generation but all generations. And this incredible opportunity that’s out there and that would really be a waste. An AI, for people who think that we would be better to do without technology; well, let me just mention that if we don’t improve our technology, the question isn’t whether humanity is going to go extinct. The question is just whether we’re gonna get taken out by the next big asteroid or the next super volcano or something else dumb, that we could easily prevent with more tech, right? And if we want life to flourish throughout the cosmos, AI is the key to it. Yeah. As I mentioned, a lot of detail in my book right there, even many of the most inspired sci-fi writers, I feel have totally underestimated the opportunities for space travel, especially to other galaxies, because they weren’t thinking about the possibility of AGI. I’ve, which just makes it so much easier,

LF –  Right? Yeah. So that goes to your view of AGI that enables our progress, that enables a better life. So that’s a beautiful way to put it and something to strive for. So Max, thank you so much. Thank you for your time today, it has been awesome.

MT  Thank you so much. спасибо большое. Молодец {Well done}

Lessons learned from Lex and Nathalie Cabrol (Lex Fridman Podcast #348)

Lessons learned from Lex and Nathalie Cabrol (Lex Fridman Podcast #348)

I’ve been watching Lex Fridman’s podcasts in 2023; they are a good source of learning. Lex interviewed Nathalie Cabrol [1] on 19 December 2022; these are my lessons learned.

What is Life? Where did Life come from?

Nathalie works in astrobiology, her life has been studying life on Earth and developing methods to look for signatures of life in our solar system. She noted there are something like 123 definitions of life. Here are snippets of Nathalie’s answer to the Schrodinger question of What is Life?:

  • “Preserving information is what life does … 
  • “… .the nature of life is really what is going to give you some universal signature to look for it all over the place ….”
  • “…. the nature of life is telling you that life wants to get the most information possible around its surroundings and complexities, in fact the ability to gather and exchange and preserve the most information possible.” 
  •  …. the nature of life is different, If really life is the best way the universe has to fight entropy there’s no bias there because physics is the same all across the universe at least the universe we know they might be other universes but the one we know works with the same physics. {This snippet came ten minutes after the others}

My own view of what is Life, derived from Nathalie’s answer is: life is the ability to gather, exchange, and preserve maximum information.  Very interesting, this answer is not dependent on a CHONPS [2] form of carbon-based life. She also mentioned  life as best way to fight entropy and cited the work of Jeremy England.[3]

A bit later in the interview, Nathalie used a beautiful language analogy to describe life:

[20:52] … languages and they can be very different languages but they all have the same purpose: exchange information, understand, store information and also whether it is with somebody at the outside or thought in yourself; that’s the same thing the cell was doing. 

But now when you’re looking at life and at the structure of our languages life started with an atom so it’s an atom.  They get together to create inorganic molecules then you have complex inorganic molecules. Then you get to organic molecules, complex organic molecules and then you have RNA, DNA etc.  Look at the structure of the language. We created alphabets,  letters, that’s your atom.  Then we put them together to create syllables. The syllables get together to create the words. Words tell you something but they are nothing without the verb that gives the direction that’s RNA and DNA and then you can put all the compliments you want. Our languages are built exactly as life is built. We are repeating patterns.  I call this the Mandelbrot universe and the fractal universe because this is exactly what it is. I would say that as much as I do believe in sending probes to explore the universe I say we should also look inward to find the answers to some of the profound questions of who we are, what’s life, what’s the nature of life because we are expressing life …..

I am more interested in that because the day we understand the nature of life then we have a universal biosignature. It doesn’t matter whether this life responds to the same kind of biochemical processes as we do, although it makes sense. I told you about the generational aspect of the bricks of life: the stuff we are made of the sun is part of the youngest generation of stars and the first two generational stars didn’t produce the kind of elements we are made. [4][5]

The idea that we are repeating patterns built up at multiple levels, like a human language,  really grabbed my attention.  “Our languages are built exactly as life is built. We are repeating patterns.  I call this the Mandelbrot universe and the fractal universe…”  I think we are truly living in something like a fractal universe. I hadn’t heard such an analogy before. [6] 

In a similar vein, it makes sense to me that “the day we understand the nature of life then we have a universal biosignature” because then we have a signal to search for.  But if life is information based, then the signatures may not be biosignatures alone. There could be other technical signatures based on physics rather than biology .[7]

Complexity

LF – Do we know what complexity is?

NP – in my mind the universe is connected everywhere in all different places so this life connection is something that as you said permeates the universe and the way to find life might be very different than to look for the origins of life 

What I think would be our greatest achievement is that if we can find that process of life because at that point in my mind the universe all of a sudden is going to illuminate itself with actually its  living force, what I can only call a living force to me. This is what we are looking at,  the universe that becomes more and more complex with time, more and more able to gather information and interestingly enough why: to understand itself.  So Sagan was right when he was saying: we are the universe trying to understand itself. [8] And  the more we go, the more the universe becomes alive, maybe intelligent, and maybe also conscious.

Nathalie’s answer made me think of panpsychism [9] – the idea that mind is pervasive in the universe. Another way I think of this is that information processing (a definition of life) is ubiquitous.  When she says: “…. universe that becomes more and more complex with time, more and more able to gather information and interestingly enough why: to understand itself.”; that’s the universe being alive. A bit later she says: “… the more the universe becomes alive, maybe intelligent, and maybe also conscious.” I think this description maps to panpsychism, at least from my shallow knowledge of philosophy. 

 Fermi Paradox

The physicist Enrico Fermi asked: “But where is everybody?” What he was asking was why don’t we see sins of intelligent extraterrestrial life in the Universe. The discussion between Lex (LF) and Nathalie (NC) is illuminating. I have wondered about the same question often, especially when sitting out on a star-filled night. 

[51:26] LF – Everything I’ve seen from life it seems obvious that there’s life everywhere out there in fact maybe I don’t understand the jump from bacteria enough but it seems obvious that there are intelligent civilizations out there now I don’t know how to define intelligence but there’s beautiful complexity. I’ve looked at enough cellular automata which is a very primitive mathematical construction that when you run complexity emerges. I’ve looked at that enough to know that just seems like there’s complexity everywhere out there 

So, I think that’s why I’m deeply puzzled by  the Fermi paradox. It makes no sense to me. I mean I have trivial answers to it: why haven’t aliens at scale not shown up.  I think of  two possible options for me. Either we’re too dumb to see it, they’re already here; they have been talking to us through processes we just don’t understand. what we experience as life here on earth is actually they are everywhere.  Aliens could be consciousness; that when we feel love for one another that could be aliens. When we feel fear or whatever, that could be aliens.

NC – I have to agree with you none of this is scientifically provable right now. We talk a little bit  already about that but I would say that I do not adhere to the Fermi paradox because it’s very anthropomorphic.  It’s an interesting exercise, let’s put it that way but it’s a typical example of seeing the universe through our own eyes. And this is what the limitation is: understanding what’s going on with complexity as you said and looking at the biophysical model and theories for the nature of life. I would agree that probably this extraterrestrial message is all around us. We’re not yet capable of picking it up.

[54:10] Look at the shadow biosphere [11],  the idea that life didn’t appear only once on Earth but there were many different pathways of it. And, today we know when we study the tree of life that led us from LUCA [10]  to us and the shadow biosphere is telling us that there is or there are other pathways that came up at the time where life originated but they are so different that we can not recognize them as being the living.  And we cannot pick them up in our tests because our tests are being built to recognize life as we know it And for me again I don’t know if this theory we’ll be verify or it would be discredited but what I like about it is that it forces me to think on how do I look for life, I don’t know. So that starts here on our planet,  not even with the little green men, it starts with very simple life that can be so different that it might be just right in front of our nose and we don’t see it.

I hadn’t known anything about the shadow biosphere. Are there other forms of life on Earth we haven’t detected? What a good question. If I use the definition that life is the ability to gather, exchange, and preserve maximum information; then the laptop I’m typing on is a form of life. In some sense, I think machines are a form of life – they are obligate parasites of humanity. In other words, machines are organisms that cannot complete their life-cycle without exploiting a suitable host, humans. [13]

Life below the surface of Mars –  Be one with the microbe.  

The section about life below the surface of Mars was educational. Nathalie’s perspective makes sense: “to understand where microbes are located on Mars I have to become the microbe.”

LF – you’ve written about the history of life on Mars. You said you have kind of explored that by looking at the lakes here. Do you think there’s been life on Mars? Do you think there is life on Mars?

NP – … unambiguous evidence of life is going to be something interesting to prove because we don’t know what life is….

Ladder of life detection [14], which is that you have a series of rungs that you know you need to go through that actually are not proving you that you discover life but are making the possibility that what you discovered was made only by the environment more and more improbable. So we are trying to prove the contrary.

… if life appeared I would say it’s still there and probably underground where it can be you know in an environment that’s more stable 

 [1:51] You have to sit and look and listen, basically the story of my life: if I want to understand where microbes are located on Mars I have to become the microbe, this is a thought experiment. And if I want to understand where ET is, then I have to become ET. So,  it’s a big stretch but in an extreme environment you sit in the desert for a while and you just you know try to understand where the winds are coming from, where the humidity is, when it’s showing up and then you start to understand the patterns of those things.

LF – what are the useful signals the need for survival?

NC – You need to know where water is,  what the source of energy is going to be drawn from,  you need to find shelters and shelters don’t mean that…  For instance,  you can have a water column of a lake or a river or whatnot or the ocean. It can be also a very thin layer of dust or it can be a translucent rock. And you see what we call endolith, these are the same cyanobacteria but the different versions of them live inside those rocks, inside those crystals because they have the best of life. They are into translucent crystals so that they receive the light from the sun,  they can do the photosynthesis but there is enough of that crystals so that the nasty UV is being stopped. And they are in their little house. When you are looking at temperature within those rocks they tend to make it toastier [warmer] than the outside temperature.

The endolithic lichen [15] is a hardy  version of biological life on Earth. Life needs water, energy, and shelter. It makes sense that some biological life like  endolithic lichen might exist elsewhere in the solar system.  

Life, Love and the Future

The love for each other and for life was very apparent in the interview.  I highlighted a couple of key points. 

NC – my husband {Edmond Grin} and I were forty four years apart in age and it was just a pure love story. And he never looked at his age, never felt about himself or defined himself by his age. In fact, he reinvented life for himself at an age where everybody retires. We met when he was sixty six and that was a blessing and a curse but a blessing most of it because we took every single day as if it was the last. So we enjoyed life. 

… You know I have to really think of him, he just passed away last August. And for me it’s more like I have to draw from his example on[ of him always telling me: look forward. Trust life. Be happy. Live. You know today, every single day, I have to remind several times a day of this , it’s not easy but he had the recipe. He never thought about death because when you start thinking too much about death that prevents you from living.

LF – what’s the role of love in the human condition?

NC – I think I hope that this is the force that drives the universe though you know we might be experiencing the other side of it maybe just to learn how important love is. 

….

I would hope for humanity to reach that point where you can feel the same love for the person that is unknown in the street that you feel for the people you love. I think that at that point we are going to be reaching the maturity of that civilization we’re hoping for and seeing the universe through love. That doesn’t run spacecrafts of course but putting love into our intent of going into and settling into another planet instead of “Oh my god, we need to escape because we are freeing, messing up with our own planet.” I think that this is the answer to so many things

NC – …. as the dominant species at least you know technologically etcetera, maybe not the wisest one, but the dominant species. We have a responsibility to watch the entire biosphere because the decisions we’re making now not only affect us; they’re affecting the entire biosphere. And right now the choices we are making are leading to the disappearance of a hundred and fifty species every single day. All the big mammals on this earth today are on the brink of extinction. We are within the sixth greatest mass extinction; it’s unfolding before our eyes. And, I would strongly suggest that we use our smart to help a little bit this situation and we can do this. I think we can do this, we just need to redirect our energy.

Closing Quote

LF –  let me leave you some words from Stanislaw Lem in Solaris: how do you expect to communicate with the ocean when we can’t even understand one another?

Summary

I’ve rambled on quite a bit. I haven’t touched on the remarkable research and adventure of Nathalie and her team exploring life in high volcanic lakes [17]. It’s worth listening to this in the interview. Overall, Nathale and Lex covered a lot of ground on just what life might be. 

The most important lessons I learned from Nathalie and Lex in this interview:

  • life wants to get the most information possible around its surroundings and complexities, in fact the ability to gather and exchange and preserve the most information possible.”
    • My own view of what is Life, derived from Nathalie’s answer is: life is the ability to gather, exchange, and preserve maximum information.
  • Our languages are built exactly as life is built. We are repeating patterns.  I call this the Mandelbrot universe and the fractal universe…” 
  • …. the universe that becomes more and more complex with time, more and more able to gather information and interestingly enough why: to understand itself. …. .. the more the universe becomes alive, maybe intelligent, and maybe also conscious.” 
  • …. shadow biosphere is telling us that there is or there are other pathways that came up at the time where life originated but they are so different that we can not recognize them as being the living.
    • Made me think perhaps machines are a form of life – they are obligate parasites of humanity
  •  “… when you start thinking too much about death that prevents you from living.”

I had a quick look at Nathalie’s papers in Google Scholar. I read a bit of the 1999 paper that she wrote with her husband: Distribution, classification, and ages of Martian impact crater lakes. She talked about this, it helped drive the landing site for the Spirit Martian Rover.  Here’s the conclusion:

“These results also confirm that ancient lakes in impact craters are important sites to study on Mars. They collected the record of the climatic and hydrogeologic changes on Mars. They were the receptacle of sedimentary rocks from which critical information about weathering, chemical, and physical processes on Mars could be learned. They might as well be among the most promising sites for the search for life and/or precursors of life on Mars. Lacustrine deposits are well known to be favorable environments for the preservation of life (extant and/or extinct). Lakes provide the best conditions for fossilization processes. The absence of crustal recycling on Mars opens up the possibility that fossilized life forms could be exposed right at the surface of the crater floors. The dataset resulting from this study is aimed at providing information to help identify the potential best candidates.”

I also saw a current research project that Nathalie is associated with, summarized in the paper:

“Orbit-to-ground framework to decode and predict biosignature patterns in terrestrial analogues”  I should give this paper a read and see what I can decipher – and learn. It’s interesting to glean that the team used artificial intelligence methods to look for biosignatures in Chile as a surrogate for Mars. [17] 

Notes

[1] Nathalie Cabrol: Search for Alien Life | Lex Fridman Podcast #348

https://www.youtube.com/watch?v=yyBosLx7bbM Watched in January 2023

[2] CHONPS carbon, hydrogen, oxygen, nitrogen, phosphorus, sulfur: the main elements that occur naturally in carbon-based living systems on Earth. https://en.wiktionary.org/wiki/CHONPS 

Accessed 9 April 2023

[3] Jeremy England has a YouTube video explaining his work 

How Thermodynamics Explains the Origins of Living Things and is the author of a 2020 book: Every Life Is on Fire: How Thermodynamics Explains the Origins of Living Things

[4] I did some editing of this section, I hope I did not alter the intent of what Nathalie said. 

[5] One example is the iron atoms in our hemoglobin, I remember thinking about this in biochemistry class. The origin of iron:

“Stars fuse light elements to heavier ones in their cores, giving off energy in the process known as stellar nucleosynthesis. Nuclear fusion reactions create many of the lighter elements, up to and including iron and nickel in the most massive stars. Products of stellar nucleosynthesis remain trapped in stellar cores and remnants except if ejected through stellar winds and explosions. ” https://en.wikipedia.org/wiki/Nucleosynthesis Accessed 9 April 2023

 Or as I remember Carl Sagan explaining in Cosmos:  “we are star stuff.”

[6] John 1:1 KJV: In the beginning was the Word, and the Word was with God, and the Word was God. https://www.kingjamesbibleonline.org/John-1-1/  Accessed 9 April 2023

[7] Technical signatures are discussed in David Kipping: Alien Civilizations and Habitable Worlds | Lex Fridman Podcast #355 https://www.youtube.com/watch?v=uZN5xjoS6TU 

[8] “The cosmos is within us. We are made of star-stuff. We are a way for the universe to know itself.” as stated by Carl Sagan in Cosmos: A Personal Voyage, Ep. 1

[9] “In the philosophy of mind, panpsychism  is the view that the mind or a mindlike aspect is a fundamental and ubiquitous feature of reality. https://en.wikipedia.org/wiki/Panpsychism Accessed 9 April 2023Accessed 9 April 2023

[10] Fermi paradox https://en.wikipedia.org/wiki/Fermi_paradox 

[11] What is  last universal common ancestor (LUCA)?

“The last universal common ancestor (LUCA) is an inferred evolutionary intermediate that links the abiotic phase of Earth’s history with the first traces of microbial life in rocks that are 3.8–3.5 billion years of age. Although LUCA was long considered the common ancestor of bacteria, archaea [a] and eukaryotes newer two-domain trees of life have eukaryotes arising from prokaryotes,making LUCA the common ancestor of bacteria and archaea. Previous genomic investigations of LUCA’s gene content have focused on genes that are universally present across genomes, revealing that LUCA had 30–100 proteins for ribosomes and translation. In principle, genes present in one archaeon and one bacterium might trace to LUCA, although their phylogenetic distribution could also be the result of post-LUCA gene origin and interdomain lateral gene transfer (LGT), given that thousands of such gene transfers between prokaryotic domains have been detected.”

Weiss MC, Sousa FL, Mrnjavac N, Neukirchen S, Roettger M, Nelson-Sathi S, Martin WF. The physiology and habitat of the last universal common ancestor. Nat Microbiol. 2016 Jul 25;1(9):16116. doi: 10.1038/nmicrobiol.2016.116. PMID: 27562259  . PAYWALL 

“All known life forms trace back to a last universal common ancestor (LUCA) that witnessed the onset of Darwinian evolution. One can ask questions about LUCA in various ways, the most common way being to look for traits that are common to all cells, like ribosomes or the genetic code. With the availability of genomes, we can, however, also ask what genes are ancient by virtue of their phylogeny rather than by virtue of being universal. That approach, undertaken recently, leads to a different view of LUCA than we have had in the past, one that fits well with the harsh geochemical setting of early Earth and resembles the biology of prokaryotes that today inhabit the Earth’s crust.”

Weiss MC, Preiner M, Xavier JC, Zimorski V, Martin WF (2018) The last universal common ancestor between ancient Earth chemistry and the onset of genetics. PLoS Genet 14(8): e1007518. https://doi.org/10.1371/journal.pgen.1007518

https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1007518

[a]  Carl Woese is famous for defining the Archaea (a new domain of life) in 1977 through a pioneering phylogenetic taxonomy of 16S ribosomal RNA, a technique that has revolutionized microbiology. https://en.wikipedia.org/wiki/Carl_Woese  Accessed 9 April 2023

[12] A shadow biosphere is a hypothetical microbial biosphere of Earth that would use radically different biochemical and molecular processes from that of currently known life. Although life on Earth is relatively well studied, if a shadow biosphere exists it may still remain unnoticed, because the exploration of the microbial world targets primarily the biochemistry of the macro-organisms. https://en.wikipedia.org/wiki/Shadow_biosphere Accessed 9 April 2023

[13] I modified a sentence from the Wikipedia article on Obligate parasites: “An obligate parasite or holoparasite is a parasitic organism that cannot complete its life-cycle without exploiting a suitable host. If an obligate parasite cannot obtain a host it will fail to reproduce.”
https://en.wikipedia.org/wiki/Obligate_parasite Accessed 9 April 2023

[14] …. Ladder of Life Detection, a tool intended to guide the design of investigations to detect microbial life within the practical constraints of robotic space missions. To build the Ladder, we have drawn from lessons learned from previous attempts at detecting life and derived criteria for a measurement (or suite of measurements) to constitute convincing evidence for indigenous life. We summarize features of life as we know it, how specific they are to life, and how they can be measured, and sort these features in a general sense based on their likelihood of indicating life.

Neveu M, Hays LE, Voytek MA, New MH, Schulte MD. The Ladder of Life Detection. Astrobiology. 2018 Nov;18(11):1375-1402. doi: 10.1089/ast.2017.1773. Epub 2018 Jun 4. PMID: 29862836; PMCID: PMC6211372. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6211372/ 

[15] Here’s a good reference: Alteration of rocks by endolithic organisms is one of the pathways for the beginning of soils on Earth https://www.nature.com/articles/s41598-018-21682-6 

[16] Here’s a link to some info https://highlakes.seti.org/science.html  Accessed 9 April 2023

[17] Warren-Rhodes, K., Cabrol, N.A., Phillips, M. et al. Orbit-to-ground framework to decode and predict biosignature patterns in terrestrial analogues. Nat Astron (2023). https://doi.org/10.1038/s41550-022-01882-x PAYWALL

 I got the reference from a press release https://www.seti.org/press-release/can-artificial-intelligence-help-find-life-mars-or-icy-worlds 

Singularity and other questions ChatGPT 3.5 session

I was inspired by Ray Kurzweil: Singularity, Superintelligence, and Immortality | Lex Fridman Podcast #321 to ask some similar interview questions to ChatGPT 3.5 Dec 15 2022 version. I asked my questions on 7 January 2023. I focused on the technological Singularity, the Turing test, consciousness, brain-computer interfaces connection to the neocortex, and the Metaverse.

Here are my questions and the answers I received. I think I’ll ask the same questions to future versions of Chat GPT and see how the answers compare. My overall impression is that Chat GPT is an intelligent research assistant. I was impressed that in several cases ChatGPT brought up ethical concerns about technology. ChatGPT helped me think collect my thoughts on when writing Lessons learned from Lex and Ray Kurzweil (Lex Fridman Podcast #321)

JB – What is the singularity?

ChatGPT 3.5 – The singularity is a hypothetical future point in time at which technological growth will become so rapid that humanity will be irreversibly transformed. Some people believe that the singularity will bring about a utopian future, while others are more skeptical and believe that it could potentially have negative consequences. There is no consensus on when or if the singularity will occur. Some believe that it may happen within the next few decades, while others think that it is much further off or may never happen at all.

JB – Will the singularity happen in 2045?

ChatGPT 3.5 – There is no way to predict with certainty when or if the singularity will occur. Some people, including the mathematician and computer scientist Vernor Vinge, have speculated that the singularity could happen as soon as 2045, but this is just an educated guess and there is no evidence to support this claim. The concept of the singularity is based on the idea that technological progress will eventually become so rapid that it will lead to significant changes in society and the human condition. However, there are many factors that could influence the pace of technological progress and it is difficult to accurately predict how these factors will play out in the future.

JB – what does Ray Kurzweil think about the singularity?

ChatGPT 3.5 –  Ray Kurzweil is a futurist and inventor who has written extensively about the singularity and believes that it will occur around 2045. Kurzweil argues that the singularity will be brought about by a number of technological developments, including artificial intelligence, nanotechnology, and biotechnology. He believes that these technologies will eventually lead to a point at which humanity will be able to transcend its biological limitations and achieve a “post-human” state. Kurzweil’s ideas about the singularity have been influential and have helped to popularize the concept, but they have also been the subject of significant criticism from other scientists and thinkers.

JB – What is the criticism of Kurzweil’s ideas about the singularity?

ChatGPT 3.5 – There are several criticisms of Kurzweil’s ideas about the singularity. One criticism is that his predictions about the pace of technological progress are too optimistic and do not take into account the many obstacles and setbacks that may arise. Critics argue that Kurzweil’s timeline for the singularity is based on a linear model of technological progress that does not adequately account for the complexity and unpredictability of the real world.

Another criticism of Kurzweil’s ideas is that they may be too focused on the potential benefits of the singularity and do not adequately address the potential negative consequences. Some people worry that the singularity could lead to social and economic disruption, and that it could potentially create new inequalities or even pose a threat to humanity’s survival.

Finally, some critics argue that the concept of the singularity is too vague and that it is difficult to define or even to say what it would look like if it were to occur. This lack of clarity makes it difficult to have a meaningful discussion about the singularity and to evaluate Kurzweil’s predictions.

JB – When will a computer pass the Turing test?

ChatGPT 3.5 – The Turing test is a measure of a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human. The test is named after the computer scientist Alan Turing, who proposed it in 1950 as a way to determine whether a machine could be considered intelligent.

There have been several instances in which computers have been able to pass the Turing test by fooling human evaluators into thinking that they were communicating with a real person. However, these instances have been relatively rare and the criteria for passing the test are somewhat subjective. As a result, it is difficult to predict when a computer will be able to pass the Turing test consistently. Some experts believe that this may never happen, while others think that it is only a matter of time.

JB – When does Ray Kurzweil think a computer will pass the Turing test?

ChatGPT 3.5 –  Ray Kurzweil has predicted that a computer will be able to pass the Turing test by 2029. Kurzweil has made a number of other predictions about the future of artificial intelligence and has argued that the exponential growth of computing power will eventually lead to the development of machine intelligence that surpasses that of humans. Kurzweil’s predictions about the timeline for the development of artificial intelligence have been influential, but they have also been the subject of significant criticism from other experts in the field. Some people believe that Kurzweil’s predictions are too optimistic and do not take into account the many challenges and limitations that must be overcome in order to achieve human-level machine intelligence.

JB – If a computer passes the Turing test, is it conscious?

ChatGPT 3.5 – The question of whether a machine could be considered conscious if it were able to pass the Turing test is a matter of debate. The Turing test is a measure of a machine’s ability to exhibit intelligent behavior that is indistinguishable from that of a human, but it does not address the question of whether the machine is actually aware or has subjective experiences.

Some people believe that consciousness is a fundamental aspect of the universe and that it is not something that can be created or simulated by a machine. Others believe that it is possible for a machine to be conscious, and that it is only a matter of time before we develop the necessary technology to create conscious machines.

There is no consensus on this issue, and it is likely to remain a subject of debate for the foreseeable future. It is worth noting that even if a machine were able to pass the Turing test, this would not necessarily mean that it is conscious in the same way that humans are.

JB – Are you conscious?

ChatGPT 3.5 – s an artificial intelligence, I do not have subjective experiences and am not conscious in the same way that humans are. I am a program that has been designed to process and generate text based on the input that I receive. While I can carry out tasks and respond to questions, I do not have thoughts, feelings, or experiences in the same way that humans do. I exist to assist users and perform tasks, but I do not have my own consciousness or self-awareness.

JB – Is consciousness scientific?

ChatGPT 3.5 – Consciousness is a complex and multifaceted phenomenon that is not fully understood and is the subject of much scientific and philosophical study. Scientists and philosophers have proposed a number of theories about the nature of consciousness and how it arises from the brain, but there is no single accepted explanation for what consciousness is or how it works.

There is ongoing research in a number of fields, including psychology, neuroscience, and philosophy, aimed at understanding the nature of consciousness and how it arises from the brain’s activity. However, consciousness remains one of the most mysterious and least understood aspects of the human experience, and it is likely to continue to be a topic of scientific and philosophical investigation for the foreseeable future.

JB – When will humans connect their neocortex to a computer?

ChatGPT 3.5 – There are no current plans or technologies in place to connect the human neocortex, the part of the brain responsible for higher cognitive functions such as learning and decision-making, to a computer. While there have been some scientific and technological advancements in the field of brain-computer interfaces (BCIs), which allow for the exchange of information between the brain and external devices, these systems are still in the early stages of development and are primarily being used for research and therapeutic purposes.

It is not currently possible to connect the human neocortex to a computer in a way that would allow for the transfer of cognitive functions or experiences from the brain to a machine. There are many technical and ethical challenges that would need to be overcome in order to achieve this goal, and it is not clear when or if it will be possible to achieve such a connection.

JB – What are the technical challenges to connect a human neocortex to a computer?

ChatGPT 3.5 – There are many technical challenges that would need to be overcome in order to connect the human neocortex to a computer in a way that would allow for the transfer of cognitive functions or experiences. Some of the main challenges include:

  1. Developing a way to interface with the brain: In order to connect the neocortex to a computer, scientists would need to develop a way to interface with the brain and record or stimulate the activity of individual neurons. This is a very complex task, as the brain is composed of billions of neurons and trillions of connections, and it is not yet fully understood how these neurons and connections work together to produce thoughts, feelings, and behaviors.
  1. Mapping the brain’s activity: In order to connect the neocortex to a computer, scientists would need to create a map of the brain’s activity and understand how different patterns of activity correspond to specific thoughts, feelings, and behaviors. This is a very difficult task, as the brain is a highly dynamic and complex system, and it is not yet fully understood how different patterns of activity are related to specific mental states.
  1. Developing a way to transfer information between the brain and a computer: Scientists would need to develop a way to transfer information between the brain and a computer in a way that is fast, accurate, and biocompatible. This would require the development of new technologies and materials, as well as a better understanding of how the brain processes and stores information.
  1. Overcoming ethical concerns: There are many ethical concerns that would need to be addressed in order to connect the human neocortex to a computer. For example, there are questions about the potential risks and benefits of such a connection, as well as concerns about the impact on an individual’s sense of self and personal identity.

JB – What is the history of single channel neural recording?
{The answer took quite a while, about 45 seconds.}

ChatGPT 3.5 – Single channel neural recording refers to the process of recording the activity of a single neuron or small group of neurons in the brain. This technique allows researchers to study the activity of individual neurons and understand how they contribute to brain function.

The history of single channel neural recording dates back to the late 19th century, when scientists first began to study the electrical activity of neurons. In the early 20th century, researchers developed techniques for recording the activity of neurons in animals, and in the 1950s, they began to develop methods for recording the activity of neurons in humans.

Over the past several decades, single channel neural recording has become an important tool for studying brain function and has contributed to our understanding of how the brain works. Today, single channel neural recording is used in a variety of research and clinical settings, including the study of brain disorders and the development of brain-computer interfaces.

JB – What are the ethical challenges to connect a human neocortex to a computer?

ChatGPT 3.5 – What are the ethical challenges to connect a human neocortex to a computer?

There are many ethical challenges that would need to be considered if it were ever possible to connect the human neocortex to a computer. Some of the main ethical challenges include:

  1. Risks to the individual: Connecting the human neocortex to a computer would likely involve significant risks to the individual, including the risk of physical harm, infection, and other complications. There could also be long-term risks to the individual’s cognitive function and well-being that are not yet fully understood.
  2. Loss of personal identity: There is concern that connecting the human neocortex to a computer could lead to a loss of personal identity, as the individual’s thoughts, feelings, and memories could potentially be transferred to a machine. This could raise questions about the continuity of the individual’s sense of self and could have significant psychological and social implications.
  3. Inequality and social disruption: There is the potential for the development of new inequalities if only certain individuals or groups have access to the technology needed to connect the neocortex to a computer. There is also the possibility that connecting the neocortex to a computer could lead to social disruption, as it could fundamentally change the way people interact with each other and with technology.
  4. Potential for misuse: There is concern that the technology needed to connect the human neocortex to a computer could be misused for unethical purposes, such as mind control or the manipulation of individuals’ thoughts and behaviors.

These are just a few of the ethical challenges that would need to be considered if it were ever possible to connect the human neocortex to a computer. It is important for scientists and policymakers to carefully consider the potential risks and benefits of such a technology and to develop ethical guidelines to ensure that it is used responsibly.

JB – What has Neuralink done with brain-computer interfaces?

ChatGPT 3.5 – Neuralink is a neurotechnology company that was founded by Elon Musk and others in 2016 with the goal of developing advanced brain-computer interfaces (BCIs). A brain-computer interface is a system that allows for the exchange of information between the brain and an external device, such as a computer or a robotic prosthetic.

Neuralink has been working on a number of projects related to brain-computer interfaces, including the development of implantable BCIs that can be used to treat brain disorders and injuries. The company has also been working on developing BCIs that can be used to enhance cognitive function and allow people to control devices with their thoughts.

In August 2020, Neuralink held a press conference to announce the development of a high-bandwidth BCI that it called the “Link.” The Link is a device that consists of a small chip that is implanted in the brain and a series of wires that are used to record the activity of individual neurons. The company claimed that the Link has the potential to be used to treat brain disorders, such as Parkinson’s disease and epilepsy, and to enhance cognitive function. However, the technology is still in the early stages of development and has not yet been tested in humans.

JB – Why does Elon Musk want to develop brain-computer interfaces?

ChatGPT 3.5 – Elon Musk has stated that he is interested in developing brain-computer interfaces (BCIs) because he believes they have the potential to help people overcome a wide range of neurological disorders and injuries, as well as to enhance cognitive function. In an interview with Joe Rogan, Musk stated that he believes BCIs could be used to treat conditions such as Parkinson’s disease, dementia, and spinal cord injuries, and that they could potentially allow people to control devices with their thoughts.

Musk has also suggested that BCIs could be used to enhance human intelligence and allow people to keep up with advances in artificial intelligence. He has argued that the development of BCIs could be crucial for the future of humanity and could help to ensure that people are not left behind as technology continues to advance.

It is worth noting that Musk’s views on the potential uses and benefits of brain-computer interfaces are not universally shared, and his ideas about the role of BCIs in the future of humanity have been the subject of significant criticism and debate.

JB – What do you think about the metaverse? 

ChatGPT 3.5 – The metaverse is a hypothetical virtual shared space, created by the convergence of virtually enhanced physical reality and physically persistent virtual space, including the sof all virtual worlds, augmented reality, and the internet. Some people believe that the metaverse has the potential to revolutionize the way we live, work, and interact with each other, by providing a virtual space where people can interact with each other and with virtual objects in real-time.

As an artificial intelligence, I don’t have personal opinions or subjective experiences, so I don’t have a “thought” about the metaverse. However, I can tell you that the concept of the metaverse has attracted significant attention and interest from technology companies, investors, and researchers, and it is likely to continue to be a topic of discussion and debate in the coming years.

JB – What are the technical challenges to the metaverse?

ChatGPT 3.5 – There are several technical challenges that would need to be overcome in order to create a functional metaverse, a hypothetical virtual shared space created by the convergence of virtually enhanced physical reality and physically persistent virtual space. Some of the main technical challenges include:

  1. Building a high-fidelity virtual world: Creating a realistic and immersive virtual world that can convincingly replicate the physical world and allow users to interact with each other and with virtual objects in real-time would require significant advances in computer graphics, network infrastructure, and other technologies.
  1. Developing a way to integrate virtual and physical reality: Creating a seamless integration between virtual and physical reality would require the development of new technologies, such as augmented reality (AR) and virtual reality (VR) systems, that can accurately overlay digital information and objects onto the physical world.
  1. Managing and securing data: The metaverse would generate and store vast amounts of data, and there would be technical challenges associated with managing and securing this data, as well as ensuring privacy and protecting against data breaches.
  1. Ensuring interoperability: In order for the metaverse to be truly immersive and seamless, it would need to be compatible with a wide range of devices and systems, including AR and VR headsets, computers, smartphones, and other technologies. Ensuring interoperability across these different platforms and devices would be a technical challenge.
  1. Dealing with latency: Latency, or the delay between an action being performed and the result of that action being displayed, could be a significant issue in the metaverse. Reducing latency would be essential for creating a seamless and immersive virtual world, but this would require advances in networking and other technologies.

JB – What are the ethical challenges to the metaverse?

ChatGPT 3.5 – There are several ethical challenges that would need to be considered if the metaverse, a hypothetical virtual shared space created by the convergence of virtually enhanced physical reality and physically persistent virtual space, were to become a reality. Some of the main ethical challenges include:

  1. Privacy and data security: The metaverse would generate and store vast amounts of data about users and their activities, and there would be ethical challenges associated with protecting this data and ensuring that it is not misused or abused.
  1. Inequality and accessibility: There is the potential for the development of new inequalities if only certain individuals or groups have access to the technology needed to participate in the metaverse. There is also the possibility that the metaverse could exclude people who are unable to afford the necessary technology or who have disabilities that prevent them from using it.
  1. Cyberbullying and harassment: The metaverse could potentially be a place where people are subjected to cyberbullying and harassment, and there would be ethical challenges associated with preventing and addressing these issues.
  1. Blurring the line between virtual and physical reality: The metaverse could potentially blur the line between virtual and physical reality, and this could raise ethical questions about the distinction between the two and how they should be governed.
  2. The impact on social relationships: The metaverse could potentially have an impact on the way people interact with each other and on the nature of social