Artwork for podcast Data Science Conversations
Perry Marshall - Why Evolutionary Biology Has Big Implications For Future AI Development
Episode 1129th September 2021 • Data Science Conversations • Damien Deighan and Philipp Diesinger
00:00:00 00:59:52

Share Episode

Shownotes

In this episode we are joined by Perry Marshall to talk about his latest scientific paper entitled “Biology Transcends the Limits of Computation”.  We also discuss his $10 million Evolution 2.0 Science Prize, which is the largest prize in the world in science currently.

His paper pushes the boundaries in the field of evolutionary biology and his science prize is driving some truly fascinating and thought provoking implications for the development of strong AI.


Transcripts

(:

This is the Data Science Conversations podcast with Damien Deighan and Dr. Phillip Diesinger. We feature cutting edge data science and AI research from the world's leading academic minds and industry practitioners so you can expand your knowledge and grow your career. This podcast is sponsored by Data Science Talent, the data science recruitment experts.

(:

Welcome to the Data Science Conversations podcast. My name is Damien Deighan, and I'm here with my co-host, Dr. Phillip Diesinger. On today's show we're back to talking about a cutting edge scientific paper, but this time with a very unique twist. The paper we are featuring has significant implications for the future development of AI. It's called Biology Transcends the Limits of Computation by Perry Marshall. By way of intro, Perry is based in Chicago and he is a very rare breed indeed in the scientific community. He's a true interdisciplinarian and, to give you a flavor of some of his incredible accomplishments, last year he co-founded the cancer and evolution symposium, which is now a working group for the American Association of Cancer Research.

(:

In 2015, he published the paradigm shifting book Evolution 2.0 Breaking the Deadlock Between Darwin and Design. That book publication led to Perry founding The Evolution 2.0 Science Prize, which is an AI and origin of life competition, which has a $10 million prize fund. It features judges from Harvard, Oxford and MIT and is currently the largest science prize in the world. In addition to his scientific trailblazing, he's one of the best known business consultants in the world. After obtaining his degree in electrical engineering, his business career began with a tech startup and, after that company was sold, Perry set up his own company in the business growth and digital marketing space. As well as writing a book on how to apply the 80:20 principle to business and marketing his reinvention of the Pareto Principle was published in the Harvard Business Review in 2018. His 80/20 Curve Tool is used by NASA's Jet Propulsion Labs at the California Institute of Technology to enhance their productivity. And finally, Perry is the author of The Ultimate Guide To Google Adwords, which is now in its sixth edition. That book is the original Google ad words bible that laid the foundations for the $100 billion pay per click industry. Perry, we're delighted to have you on the show. Thank you so much for joining us.

(:

It's great to be here talking about AI to a bunch of data scientists. I'm delighted to be here. That's great.

(:

Good. Excellent. So if we start with your personal journey, could you tell us how you ended up pivoting all the way from Google ad words, the world's leading authority on that, into pushing the boundaries forward in the field of evolutionary biology?

(:

In the late 90’s, I was working in an industrial networking company, and I was their marketing manager and I had struggled mightily as a sales guy. So I was in engineering, I got laid off from my job. I ended up getting thrown into sales and thrown in the drink and having to swim and I discovered marketing. And along the way, I discovered that the best thing you can do if you're in marketing is be an authority on a subject. And I was given an opportunity to write an ethernet book. And so I'm like well, that sounds like a lot of work, but it's probably a pretty good idea. So I wrote this ethernet book. Fast forward a couple of years, I'm doing marketing consulting for IT companies and Google comes along and I started figuring out how to use Google's advertising system.

(:

Google went supernova and, you know, in 2003 they were just this weird company. They were not the 800 pound gorilla of the internet, but they got some things really, really right early on and they started taking over. And so the key to Google advertising was it's essentially a Darwinian system where you bid on any term in the English language and you write an ad and you pay money when somebody clicks on your ad and the more effective your ad is, the less you have to pay and the more customers you get for less money. And so it's this giant meritocracy, and it works exactly the same way as Darwinian evolution, except that the participants are all intelligent. And I got in this argument with my brother in 2004, because he was a missionary in China and he was losing his faith. We were both pastor’s kids. I went to church every Sunday and I was a Christian. And Brian had been totally on board with all of that until he started asking a lot of hard questions. And after four years in China, he didn't really believe in any of this stuff anymore and he was reading lots of intellectual books and science books and everything and we got into this argument. And I go Brian, look at the hand at the end of your arm. I said, this is a nice piece of engineering. And I'm an engineer, I should know. You don't believe this is an accumulation of random accidents, do you? And he goes, hold on. And he just came right back with kind of a standard Neo Darwinian answer. All you need is random mutation, natural selection and enough time and you'll get a hand and you don't need a designer or anything. And I listened to that and I thought, okay, I don't have any problem with the idea of evolution, I'm fine with that. But what Brian is telling me is that there's nothing purposeful in it, that you're going to get purpose out of no purpose if you just turn the crank long enough. And I could not think of anything I had ever learned in engineering that would verify that. Because I've been in all kinds of industries, I've designed all kinds of things. And I thought, well I know a bunch of biologists would agree with him and not me. So what is it that I don't know? Or what is it that they don't know? Like, what's the disconnect here and I went home from China, determined to get to the bottom of that. You know, when you say, how do you get from Google ads to evolutionary biology?

(:

One of the keys linking these two worlds was the very Darwinian, or almost Darwinian nature, of Google's advertising system. What I found when I went deep into evolutionary biology was that all cells are intelligent. All cells have agency, all cells have what AI people would call swarm intelligence and Darwinian evolution is not random and blind and purposeless. It is purposeful. It is driven by the agency of the cells, and it actually works a lot more like Google ad words to put it in a different context than most people would have ever realised. That it's all intelligent agents competing for resources. And that, what you see on an advertising system like Google, is just a fractal aspect of how the principles that run the entire world, the entire ecosystem and business is just a subset of that.

(:

You mentioned the fractal aspect. What did you mean by that?

(:

Evolution is chaos resolved by intent. We could have five plumbing companies in Springfield, Arkansas competing for clicks on Google, or we could have a Petri dish of bacteria competing to eat the sugar or any other competitive. We could have the super bowl or soccer matches and they all run on the same set of principles. There's a finite amount of resources. The more ingenuity you have, the more you can do with the resources. And there is the infinite ability to produce more beings to consume more resources. And so there's always an 80:20 relationship between the haves and the have-nots. And the key is ingenuity. How cleverly and resourcefully can you use the resources that are available to you? And so business is biology and biology is business. And we could say the same thing about technology. They all run on the same set of principles. They all run on the same laws of physics, the same laws of chemistry, the same information theory. What I'm really trying to do with my interdisciplinary project is connect the dots between all of these disciplines so that everybody can see the other disciplines clearly for what they are.

(:

If you are comparing evolution and Google ads they are a bunch of things that look very different, obviously, right? So the timescales are very different. For Google ads you mentioned engineers, yes, so we have very intelligent players who are actually designing things, yeah. Versus on the side of evolution, you have a more statistical process, let's say it's happening on a very broad scale and over a long timescale, but individually the steps can be quite random. So there's a lot of differences there.

(:

Randomness and evolution is overrated. Evolution is not random events, it's response to random events. Dennis Noble is one of my prize judges at Oxford University, and he has a term harnessing stochasticity. And what that term means is that a system is given random inputs and it grabs it. It holds some things fixed and allows a much smaller number of variables to mutate until it finds a solution. So your immune system does this. It's kind of like a one-armed bandit poker machine where, you know, you got the strawberries and the bananas and the cherries spinning and what the immune system does is it doesn't quite know what antigen is going to fight an invader, but it knows most of it. And so it will hold most of it the same and allow some of it to vary randomly. And so it's doing exactly what we do when we build genetic algorithms.

(:

The important thing about this is, most of the system is not random. A small part of the system is random and the agency of the cell is deciding what to hold constant and what to vary. This is how people use genetic algorithms. Like with genetic algorithms, you can't just turn it on and walk away. You have to give it goals. You have to give it fitness criteria. You have to give it very narrow constraints otherwise you get into common tutorial explosion, and then you let the system run, and then it finds a solution for you that there may not have been an analytical solution for. So in most textbooks and most evolution books, the mutations of evolution are characterised as random but, for the most part, they're not. They are calculated predictions that the organism is making. So this is true in cancer. This is true when bacteria fighting antibiotics, it's true in your immune system. And it's true in the process of one species evolving into another.

(:

But isn't one of the key, and I'm talking as a naive ignorant physicist here of course, but isn't one of the key concepts of evolution that you basically have some sort of survival of the fittest, where basically you're assuming a random pool of mutations basically, and the bacteria that adapted the best, they are the ones that survive and the other ones they basically die away. Right? So you have a kind of a survival bias and if you do that on long timescales yeah, in vast amounts of space, like the oceans must be huge for bacteria, for instance, right? At some point, all of the oceans of our planet, basically being one big cell, if you want. So it's all of these molecules floating around in there and reacting and meeting randomly on very long timescales, obviously. So isn't the core principle of evolution that you have a lot of randomness basically, and then just the fittest survive.

(:

Well, that's what everybody tells you. And I found out that's not true. There's no way it can be true. It's much more directional than that. So a typical bacteria has a million base pairs of DNA and all you have to do to prove it to yourself is ask yourself how many combinations can you make four to the power of a million? How much is that? And how many combinations are there in human DNA, which has a 3 billion base pairs. So for the power of 3 billion, how big is that number? Well, if you have more than about 50 bytes of data, you cannot derive it from any random process in the history of the universe. You hit combinatorial explosion. So what's actually going on with bacteria in humans and monkeys and birds and fish and everything is that there are very specific processes. And in my book Evolution 2.0, I call them the Swiss army knife. Organisms exchange genes with each other, organisms rearrange genes, or organisms do symbiotic mergers with each other. They do sexual hybridization with each other. They do epigenetic changes, which adjust the expression of genes without adjusting the genes themselves. And all of these systems are very contextual. And so, for example, if a woman smokes, she will pass asthma to her daughter through the egg epigenetically who will pass asthma to her granddaughter epigenetically. And the granddaughter will have asthma because the mother's body was trying to adjust to the smoking. And this is actively, her body actively reprograms her own egg to equip her offspring to live more successfully in a hostile environment. This kind of thing is going on 24/7, 365 for 3 billion years and this is why we have evolution. So in other words, evolution is driven by the intelligence of cells, not merely a random process. Because again, as I said, if you have more than about 50 bytes of data, there's not enough random processes in the history of the universe to generate all those combinations. And so evolution operates on a much, much, much narrower set of constraints than pure randomness and selection alone. And so what I discovered was there was 70 years of literature that most textbooks and most evolution books were not talking about, but it's work from people like Lynn Margulis, Barbara McClintock, James Shapiro, Waddington, where the changes in genes are very directional and they are controlled by the organism. So this is actually much, much more sophisticated than most people have ever been told.

(:

Are you saying then that natural selection has no role to play or it has some role to play?

(:

Oh, it certainly has a role to play. It has the exact same role as the playoffs have in football or basketball. If you say, how did the Cubs win the world series? You go, playoffs. And people go, oh, they won the playoffs. And you go, okay. But if you really want to know how they won the world series, you have to ask, well, how did they pick their team members, and what was their offensive strategy and what was their defensive strategy? And like, what happened in these games and you know, who hit a home run? It's just like that in biology, natural selection is an outcome, it's not an explanation.

(:

Okay, so if we look at the Cambrian explosion, for example, that was a result of these methods you've described - transposition, epigenetics that gets it to a certain stage and then natural selection takes over?

(:

Then natural selection sifts out the winners from the losers, which it's very good at doing right. So, like on Google advertising, natural selection is what ads do people click on. And what do people go buy? There's natural selection going on on Amazon all the time. The way you win the game in internet marketing is you deliberately harness natural selection to give you a statistically significant result, as fast as humanly possible. So you can kill your not very good ideas as fast as possible, and funnel all of your energy into the good ideas. So natural selection is always important. It's just that it's the last step, it's not the generator of creativity.

(:

So in your paper, you're basically taking this one step further. You're talking about cognition. I think this is something, another topic about biological systems and now maybe artificial intelligence have some things in common. Could you talk a little bit about that?

(:

Yeah. So to give you an oversimplified definition, cognition is what a lot of us would call strong AI. It is the ability to harness unlimited associative learning in almost unlimited number of contexts. There's a paper by James Shapiro called All Living Cells Are Cognitive that got published last year. And we have known since the beginning of the 20th century, that all cells are social. They communicate with each other. Most life forms live in colonies, including bacteria. Bacteria and other cells genetically change their own genetics to differentiate into different tasks. So even a colony of single cell organisms will function like a multicell organism. And Barbara McClintock determined that when you subject cells and plants and organisms to an infinite number of situations, they will respond with an infinite number of possible responses that are not predictable. And so she actually discovered there are certain things that organisms will respond with, essentially an algorithmic response.

(:

If you subject cells to heat or starvation or different things, they will respond in predictable ways. If you subject them to things that they've never been subjected to in the history of their lineage, they will still respond, but in an unforeseeable manner and they will somehow innovate. And so she discovered this with corn in the 1940s. She discovered if she damaged corn chromosomes so that the plant could not reproduce, the plant would start rearranging its chromosomes and it would find code from other chromosomes that would repair the code that was destroyed by radiation in her experiment and the plant would go on to reproduce. And this was one of the most important discoveries in the history of biology and her colleagues thought she was crazy. Half of them laughed at her. The other half were mad that she presented this at Cold Spring Harbor in New York in 1951. And she was basically driven underground and she stopped publishing her research for about 20 years, but she won the Nobel Prize for discovering transposition in 1983. And what she had really done was she was the first person to trigger an evolutionary adaptation, watch the adaptation happen, and then figure out what had happened genetically. And she did this before the discovery of DNA. And so what she really discovered was that even a corn plant has an intelligence and an ability to rearrange its own DNA and do so purposefully and contextually.

(:

So maybe coming back a little bit to the term cognition now. When you say, you mean a strong AI basically with it. How would you precisely define what criteria would assist them, have to show to qualify for an active cognition capability?

(:

Cognition is the ability to sense and respond to the environment in a non algorithmic way. If I do something that's happened to the system before it will respond, doing what was successful before. But if I do something completely unexpected, like let's say, a giant meteor comes through Damien's ceiling, five seconds from now and lands on his floor. This has never happened to him before and he is going to respond contextually. He might check and see if the kid is okay in the next bedroom. He might see if it destroyed his hard drive. He might stick around and fix the ceiling. He might move the family into a hotel. He can do any number of things to a completely unpredictable situation and do it in such a way as to still achieve a goal. It's like, well I have a really important dinner tonight, and even though the meteor came through my ceiling, I still made it to dinner. That's what I mean by cognition. An algorithm does not do that.

(:

Do you think, like, would a Tesla qualify?

(:

A Tesla has no cognition. There are no computers that have cognition because computers do not reason inductively. So what most AI is doing currently is, it is using huge amounts of data, oftentimes vast amounts of data, to deductively analyze a huge number of situations, and then formulate a response to each one and then train the system to formulate it. But all of this has been done deductively. And if you have terabytes or petabytes of data, it's not very hard to win a go game or a chess game or have a self-driving car, but the self-driving car is not intelligent because it still can't do inductive reasoning. So inductive reasoning is really at the heart of this. Inductive reasoning is the ability to make an educated guess when a calculation is not possible.

(:

Sensing and responding to the environment is obviously a very low threshold, yeah, that's a low hurdle to take, any feedback system can do that. My web browser can do that, they are always collecting code in the web browser. You know, it's while we are talking, basically this is happening in my computer. And also we don't doubt that Damien has cognition, I think we agree on that yeah now?

(:

A little bit.

(:

I think with the training process, I'm not too sure to be honest. I think also as humans, you know, we get out of the womb very early. We still have to learn a lot and adapt to the environment. We process a ton of data in the first years of our lives, yeah, and we build those neural connections. I think that's similar to training, you know, complex neural network in some ways, at least. Yeah. I agree, inductive logic dealing with new situations from knowledge that you have like an abstract layer of information, if you want, and utilising that to deal with new situations, that's something where AI is not yet that I can see. So if you include that into the definition of cognition, I would agree that maybe a self-driving vehicle is not there yet. But, coming back to what you said earlier, yeah. So do you think then that cells or bacteria have that level of cognition that a Tesla wouldn't have?

(:

Yes, absolutely. In fact, all living cells have cognition. Now I'm not saying they're as smart as us. They may have only a very narrow range of what they can do, but they do have it. There's a wonderful Ted Talk by Bonnie Bassler called How Bacteria Talk. And she explains how, when you're coming down with something, I feel it, right? Like you go outside on a cold day and you come back in and you start feeling that in your body. Well, those bacteria have been in your body for several weeks and they've been quietly multiplying and they have this thing called quorum sensing where they talk to each other and they go, so is there enough of us to attack? And there's a point at which the answer is, yes, let's go. And then they start consuming resources and going out into your body. And they do this by talking to each other and they have chemical signals for words like us and them, and this is how signaling works in fireflies where you have symbiotic bacteria making the glow happen. And she goes into this very sophisticated example of fish in the ocean that have these symbiotic luminescent bacteria that keeps them from getting caught by predators. And it's incredibly sophisticated. And all biology is it's communication networks all the way down to the sub cellular level. And it's all contextual. So if you put bacteria in a situation they've never been in before, you probably kill most of them, but a few of them will adapt. And they're not just doing this by chance. If you want an analogy, bacteria are like entrepreneurs. If there's a pandemic and it puts most of the restaurants out of business, it won't put all of the restaurants out of business because somebody is going to come up with something like, well, let's put delivery guys on a bicycle. Let's use Uber eats. Let's try this. Let's try that. And since there's an infinite number of things they can try. Computers by definition, like by turning machine definition only deduce they only reason from the general to the specific, they cannot reason from the specific to the general. Living things reason from the specific to the general, it's actually implicit in any assigning of symbols to meaning, okay? So you've got a living organism, it has a cell wall and it is sensing and interpreting what's outside the cell wall. That is by definition, inductive reasoning. If it has not been told in advance what it is or how it works. Okay. So babies use inductive reasoning to eventually figure out object permanence.

(:

They don't know that objects are permanent when they're six months old, but they know by the time they're four. That is genuinely inductive reasoning. They'll figure it out without being told. So there's this huge gap between living things and non-living things. And the gap is induction. Now what I say in the paper is, it was actually a pretty profound realisation. I was trying to decide, how am I going to define induction? And I realised, well, it's anything that's not deduction. And the dividing line between induction and deduction is identical to the halting problem. Is the computer going to arrive at a solution? The halting problem itself is proof. The fact that we have it, the fact that we ask the question is proof that we have inductive reasoning and computers don't. Computers only do deductive reasoning. Biology does inductive reasoning. Therefore biology is not a computer. Therefore what biology does is not computable. It has immense implications for AI. It has immense implications for biology. It has implications for philosophy, so we can go anywhere you want with that.

(:

If you look at something like consciousness. yeah. So some systems we know, pretty sure are conscious, they are self aware also, yeah. So, as for instance, there's other systems that are not, if you pick up a rock or something from the street, you know, we can all agree that this is not the case. But then you have a lot happening in between. You have a bacteria, you have cells, you have small mammals, you have insects and so on, yeah. I think at the moment, from the research, what you believe is that there's different levels of consciousness. So it's not just a switch it's on or off, but there's different levels. So it gradually develops basically. And you can have very conscientious people and not so conscientious people. And you can have also different levels of self awareness and consciousness within animals for instance.

(:

Would you say that computers would never be able to develop consciousness? If we go back to the example with the self-driving vehicles, and so at the moment, of course, we are only training it to survive in traffic. Yeah. So we're exposing it to data, millions of different situations and to learn how to handle it. But somebody could start exposing that same neural network then to other situations and to learn other things that it cannot do at the moment and then expand and expand and expand over hundreds of years or so. And at some point it has all of these different systems like we do in our brain, which are specialised. And maybe at some point, you know, data comes together on a meta level, but it's a higher abstraction. And it starts to realise that thing in the mirror that's me, you know, or something like that. Is it what you're saying, is it that this is impossible, that will never happen. It's just exclusive to biological systems or is it that you're saying we are not there yet, and it's still a process?

(:

So one of the main points of the paper, in fact, it's right in the abstract, is we don't have any examples in any literature, anywhere of chemicals producing codes. And we don't have any examples, anywhere of codes producing cognition or consciousness. Like I think just for today, I'm willing to say that cognition is roughly similar to consciousness or maybe, you know, maybe consciousness as we typically speak of it is a subset of cognition. Like it's a particular form of cognition. And, computation and cognition or computation and consciousness, as far as I can tell, are two completely different things. I don't believe any computer has an iota of consciousness. I don't see any evidence that they do. I'm an electrical engineer. I know what a computer is. I know what it's made of.

(:

I would say emerging properties. Right. So knowing what it's made of, I mean, we also know what humans are made of, right, and they have consciousness.

(:

Well, okay. That's, that's true. And so the proof is in the engineering, right. Engineering is really the gold standard of science, not peer review. Can you build it and does it work and what does it actually do? I don't know of any AI that acts like it's conscious. Like, well, if it walks like a duck and talks like a duck and quacks like a duck, then it's a duck.

(:

If you look into what happened in physics in the last decade or so, there's a lot of interesting developments when it comes to information theory or principles that are derived from information theory, right? So everything that's happening around black holes and so on, or quantum information and so on, so forth. So we have very strong principles. They are basically that seem to be, almost equal to laws of physics that we have discovered already. We see the universe more now as this huge box of vast emerging properties. Yeah. So it started with just radiation and then out of nothing, basically it formed quarks, which, you know, it's already hard to understand for us. And then leading to protons and neutrons and simple atoms, and then, later, you know, bigger atoms and then molecules and so on life. And so it's amazing how many hierarchies of emerging properties there are, right. Even now to consciousness and so on. In a way you can see, you know, if you think about the multiverse, our universe could be the one that has insane emerging, degrees of emerging properties in a way. And one of the consequences, I think if you look into the AI sector now, is that people believe that any information processing system will develop consciousness or these capabilities when it becomes complex enough and abstract enough, yeah. I agree that we don't have examples yet that it exists yet. But the question is, do you think we could get there in 50 years or 150 years, or you think this is categorically just impossible because there's like a little secret ingredient that's just missing that the biological systems have that the machines don't have?

(:

So first of all, I'm all for emergent properties. I totally get what they are. If you can demonstrate that snowflakes emerge from cold water in your refrigerator, like you proved it, right? Like the proof is in the engineering. If you can build an algorithm or a computer of which consciousness is an emergent property, more power to you and my investment group wants to buy that from you. And like, let's, let's go to town. Okay. So I am all for it. Now, I believe that cognition is, so to speak, at a right angle to computation. My paper's title is Biology Transcends the Limits of Computation and what it's saying is that what biology does is, by definition, impossible to compute because inductive reasoning is not computable. Inductive reasoning, wherever it comes from, is not derivable from mathematics. It transcends mathematics. This goes to fundamental definitions of mathematics itself. So I talked about the halting problem. The halting problem is Alan Turing proved that you don't know if a program is going to crash or go in an infinite loop or arrive at a solution without running the program. And there's not a generalised formula for looking at a software program and knowing what it's going to do.

(:

You can guess. If you have a lot of experience, you could make a very educated guess, and that would be inductive reason, but there's no way to know for sure until you've actually run the program. And so, the proof is really in the pudding and it may very well be that consciousness is a product of a certain quantum state or certain interactions of certain kinds of molecules. It could be that quantum computers could be conscious. I don't know. All of those are maybe possible, but Boolean logic in computation on transistor gates all by itself, I don't believe is ever going to get you to consciousness.

(:

Don't you think like inductive reasoning could be just an emerging property out of a complex information processing system?

(:

Well it might be, and I guess what you could say is my investment group is looking if somebody can get that emergent property to happen. Okay, whatever you did to make that happen, we want to buy the patent. We want to own this and we want to commercialise it. Really you could say that's what the prize is looking for, an emergent property of like chemicals, emerging codes or codes emerging cognition. Either one of those, oh that would totally get our attention. And that's one of the reasons that I put this prize together because I realised if anybody solves any of these really fundamental problems in biology, they will have made a breakthrough in technology that is bigger than the transistor, as big as E = mc2 or anything like that. And so it's kinda like, you know, what if in 1895, somebody came along and they said, Hey, you know what, there's some anomalies in physics and there's some stuff that's just not making sense, and I think Newtonian physics is breaking down, and if we can replace it with a better theory, that explains A, B and C, we want to give that guy a bunch of money and we want to own the patents. And it would be like, if somebody did that, and then Einstein comes along in 1905 and says, yeah, I got the theory of relativity. And then the organisation says, all right, so we own the patent on nuclear power. That's what Natural Code LLC is doing.

(:

I think in 1895 no one had already figured out special relativity actually, a mathematician before Einstein.

(:

There are always precursors to these things, they never just come out of the blue. There is always indication and people ignore it, ignore it, ignore it, ignore it and then somebody finally nails it and the world changes.

(:

Can you talk more about the prize? What are the participation criteria? Why did you set it up? What's the motivation behind it?

(:

I have several motivations. In 2005, I was listening to an NPR station and they were interviewing Richard Dawkins, the world's most famous atheist, and he wrote all these evolution books. And somebody called on the phone and they said, Professor Dawkins, where did life come from? And he goes, it was a happy chemical accident and then he just went on to the next caller. And I listened to that and I thought, did I hear that correctly? Did a professor at Oxford University just tell me that life is a happy chemical accident. Like I was aghast. And I thought if famous public intellectuals are getting away with saying that kind of nonsense, we have a serious problem in biology, a serious problem in science and a serious problem at Oxford University and a serious problem at NBR. I'm tired of hearing these silly, just so explanations for some of the most profound, scientific questions of all time and I got to thinking, you know, how do you, how do you kind of force an honest answer? And so I put together this prize and I thought, okay, let's get away from happy chemical accidents. That's just an irresponsible way of describing it. Let's say it is an emergent property. What would happen if somebody figured this out? What would happen if somebody figured out, like how a corn plant knows how to rearrange its own DNA, even though the DNA built the corn plant in the first place. I mean, that's mind bending how that works. What if somebody figured out how that worked? Well, it would revolutionise AI. It would revolutionise technology. It would give us a completely different understanding of cancer and infectious diseases and all kinds of things. It would probably completely turn medicine upside down.

(:

So why don't I go raise some money, and I did. It took a long time. It was very hard to explain this to people, let me tell you. Like, what is this crazy thing that you're trying to do? But I eventually raised the money and we announced it at The Royal Society in Great Britain two years ago. So,here's what we're looking for. An emergent communication system. And the communication system has to follow the classical definitions from electrical engineering, which is encoder - message - decoder, and the three have to correspond to each other. So, if I send Damien a text message, I type the stuff into my phone, I press send, it gets encoded into ones and zeros, and the same text appears on his phone and I can match these three things. And that is the definition of communication. And what we said is, you need to come up with some chemical process that will produce a digital communication system without cheating and the communication system simply needs to be capable of representing 32 states or like two to the power of five, so five bits of data in all of the states that are available in those. Like the genetic code handle 64, we said, we need something that's at least half as sophisticated as the genetic code and it has to be digital data. And, if you can produce a system that will do that, then I think you will have inadvertently invoked the principles of emergence that caused life to form. And if you do that, we will pay for the patents. When the patent is granted, we write you a cheque for $10 million and we partner you in the company so you can continue to enjoy the profits from the commercialisation process. And probably, whatever you did is a Nobel Prize.

(:

So when you say it would use these systems, what would qualify for that? Is it, you don't want to see a lot of engineering. It sounds to me like it's more like a soup of molecules that if you shake it a little bit, then stuff starts happening.

(:

Well. So what I like to say is, if you pour chemicals in your bathtub and you got one end of the bathtub to send messages to the other end and they correspond, and it's clearly a code, like me and Damien sending text messages, then you've done this. And so you need an encoding table and a decoding table. So the genetic code, like the standard basic genetic code that you find in any biology textbook, where GGG is instructions to make glycine and AAA is instructions to make lysine and you put that on a table. If this system will produce a table like that, because DNA gets transcribed into RNA and translated into amino acids and you make tables to show what the results are. If you can make a similar table for this thing that you made in your bathtub, you win the prize.

(:

So what you want to see is some biochemical system basically going on, right?

(:

Yes. And it could be biochemical, it could be silicon. We don't really care what it is, as long as you can extrapolate those criteria out of it, you win the money.

(:

It sounds to me a little bit like you need a really big bathtub and a hundred million years.

(:

In the traditional way of viewing biology, that would be true. But that usually assumes that it's like luck.

(:

Biochemistry can be pretty slow, right? So there's some, when we do DNA tests or whatever, right. We have sometimes limits that are just pure laws of physics basically or reaction rates.

(:

Okay, so yes, you're right. But what really shaped my thinking on this was when I saw how fast bacteria can formulate completely novel responses to difficult situations. Bacteria can develop resistance to antibiotics in 20 minutes, they are creating information when they do this. I think there's some principle that they are doing that is the same principle that causes life to emerge. So here's a thought experiment. What if cognition and consciousness are the result of a particular arrangement of molecules in quantum states? And what if I put these molecules together in a certain arrangement and I suddenly pops out a consciousness field. Now I realise I'm talking science fiction, you'll have to forgive me here. But what if suddenly, as a result of that quantum non-locality, the whole can act on its own parts on behalf of the whole. And let's say it's just one molecule, but if I have one molecule that possesses cognition, it can start building something. I think there's a principle, something resembling that in biology that we haven't discovered yet. I don't think it's waiting a hundred million years in a hundred trillion litres of water for something to accidentally happen. I think that's feudal.

(:

Well, yeah. I mean, if you go back to the example with the bacteria, 20 minutes for them means like, I don't know, five generations or something like that. Right? If you compare it to human lifespans, that would be a lot of time. We've achieved a lot in the last five generations. And with bacteria, you also have already some very complex kind of system that you start with in the first place. I mean, I'm not a biochemist, so I cannot really talk to this. But what I know about biochemistry is that there are these limitations such as to have a lot of molecules in the bathtub, so it's 10 to the 23 or 24, something like that. So that's, that's a lot but if you really rely on randomness, then you need a lot of time probably right. Somebody could just crunch the numbers to tell us. So for the prize, like how would somebody participate?

(:

If you could build it once you could build it twice, if it works, it should be repeatable and this is likely a process patent. What apparatus did you hook up to your bathtub to get these chemicals to do this and whatever that is, it's really significant. Now I had a very funny conversation once with Nasr Batho, who is the Director of Investments of EMAAR Corporation, which is the company that built the Burj Khalifa. And he watched my presentation and I was pitching for investors in 2016, and he goes, So Perry, this is really fascinating, totally tracking with you on this, I love what you're doing, I can't invest in this because my charter is real estate, and like, this is not real estate, so sorry. But he goes, I know who's going to win this. And I go, you do? And he goes, yeah. He goes, it's not going to be some scientist in a white lab coat at an American university. He goes, I think it'll be like some kid in some artistic country in Italy or Sweden who goes to Montessori School when they're like 9 or 14 years old. I think that's who's going to figure this out. And when they figure this out, everybody will say, well, why didn't we think of that? Like, I don't know. I guess it's because you weren't in Montessori School in Italy. And I just thought, you know, I can totally believe that it's going to come from an outsider.

(:

I've had very notable scientists say, Hey, I think I know how to solve that. I was on a podcast thing with Lee Cronin, who is at the University of Glasgow in Scotland and he is a rock star chemist and he runs one of the biggest chemistry labs in academia. And, he goes, I'm going to win that thing and we talked about it and he endorsed the prize and I've had some very, very notable scientists come to me and go, I think we're going to solve that. And I mean, I'm open to wherever it comes from. And most importantly, I love the conversations that it is stimulating because they, it is forcing people to ask questions that they weren't asking before. I think the origin of life field has been in a rut frankly for about 30 years. And I don't know that we really need to get into that. We gotta have fresh thinking. We got to have people from multiple disciplines asking this question. We got to get them talking to each other. And we got to, we got to solve these, these intractable problems.

(:

My paper talks about cancer. And there, there is a point in a human body where a regular pancreatic cell says, I think I want to be a rebel. And I think I want to start a movement. And it does. And now you have pancreatic cancer. Well, there's a moment there where a decision is getting made. What is that decision? I have a hunch that that decision has something to do with the origin of life itself. It's the ability to make a decision. And so I think evolution, origin of life, consciousness, cognition and AI are one question. They're not six questions, they're one question. Where does the ability to make a choice come from? What is a choice and where does it come from? I don't know. I think, I think it's one of the most important questions in all of science and, if we figure it out, we're in a whole new world. It's a scary world, by the way I'll admit, but it's a new world.

(:

Yeah, that sounds good. I mean, those are super interesting topics so we're living in a good time. And, I think what artificial intelligence is bringing to the discussion is just a new perspective that wasn't available before. That's, if you talk to [inaudible] about this, I mean, I think he's like the Einstein situation that you described before. He's already like two or three steps ahead of all of us, but I think he would also argue that probably it is some sort of emerging property and you will be able to see it in maybe 50 years or so. And then, at that point, hopefully we can dissect a little bit more and understand where it's coming from.

(:

It will bring a lot of new problems that we'll need to solve. It's like splitting the atom, you know, there's before and there's after, and there's no going back.

(:

Have you had any interesting entries to the prize yet that nearly won, or?

(:

We've gotten a number of interesting entries and if you go to Naturalcode.org and there's a link and you can go look at past submissions. It's this mish-mash of this whole range of, you know, from the ridiculous to the sublime. And I just really love where the educated conversations are going.

(:

So it's not too late for people to get together data scientists with their friends and have a crack at the prize.

(:

No, and in fact, what I would say to data scientists is, you know, whether you think you can win the prize or not, I think you should all get a lot more educated about biology. I think there's a lot, I think almost every problem that Silicon Valley is trying to solve has already been solved in the cell. I would encourage the data scientists to read my book Evolution 2.0, and see if it doesn't open like whole worlds to you and go start watching YouTube videos and, you know, listen to our podcast. We have Evolution 2.0 podcast. We're always talking to these interesting people. I think if you go, if you're like, well I got this little side hobby of just kind of going down the evolutionary biology rabbit hole and learning about swarm intelligence and cells. Or, you know, another person I think your people would be really interested in is Michael Levin, he has a Ted Talk, just search Michael Levin Ted Talk. It will blow your mind what this guy is doing. There is no way a data scientist couldn't get totally fascinated with this stuff. And the nice part is like, you don't have to get, be good at it anytime soon, you just have to think it's interesting and it will, it will feed you ideas that you would've never thought of, you know, reading a computer science book.

(:

Fantastic Perry. Yeah. If people want to hear more Perry about what you're doing, either about the prize or what you're up to, where should they go to find out more?

(:

Go to evo2.org and get on our mailing list. Scroll down and you get three free chapters of Evolution, 2.0 and at evo2.org you'll see there's links to the prize, the YouTube channel, the podcast, all of that stuff is in one place. You can follow us on social media. And, you know, I had no idea how fascinating this was going to be. When I had that argument with my brother in China, I was scared to death. I had no idea where this was going to take me and I just kind of leaped into the void. It has got, this has gotten so interesting. I mean, how is it possible that an electrical engineer with a bachelor's degree from the University of Nebraska is mixing it up with all these world-class scientists at all these major universities and getting phone calls like, Hey Perry, I think I know how to solve your prize. And, just in general, I want to encourage your listeners, you need to indulge your curiosities and when you're fascinated with stuff like whatever it is, it is not self-indulgent for you to go to pursue that. I think you should have hobbies. You should have sight interests, you should read stuff on crazy topics. You should do stuff that doesn't have any obvious connection to your regular work. It will enhance your career so much. I can't even tell you.

(:

Hmm, words of wisdom indeed. So sadly that brings today's marvelous episode to a close. Really, really fascinating conversation Perry. Thank you so much for joining us, it's been great.

(:

Thank you, Phillip. Thank you, Damien. It's an honour and great to be with your listeners today.

(:

Thank you also to my co-host, Phillip Diesinger for his usual excellent questions. And of course, thanks to you guys for listening into this episode. Do leave us a review on your favorite podcasting platform. And if you loved this episode, you'll definitely enjoy our previous two episodes on the XPRIZE and the IBM Watson prize. So we'll look forward to having you with us on the next podcast, take care, and we will catch you guys soon.

Links