I knew Blaise when he was dropping out of the Princeton PhD physics program because he had invented foveated rendering and a wicked fast general purpose mipmapped UI. I’ve met some very bright people in my life but Blaise is at the very top of that list. I’m proud of him and glad to see his mind is running free at Google. Excellent podcast. No way to walk away from this conversation without something to think about.
The most interesting talk I've heard in the last 20 years. So many things Blaise speaks about observing in his simulations have been speculated on in the origin of life community for decades (such as life-like autocatalytic chemistries being attractors in chemical space). We live in interesting times!
What a great conversation - thanks a million for posting! I would love a discussion between Steven Wolfram and Blaise Agüera y Arcas moderated by Mr. Carroll.....
Great conversation between two great minds. One note on Kolmogorov complexity being easy to approximate: this is actually a concept most people get partially wrong since we can only really put an upper bound on the Kolmogorov complexity (say of a string, unless it’s a very simple one) so using techniques like Lempel-Ziv or Levin’s KT are really more of upper bounds. That being said I am guilty of that myself as I often use the word approximation when talking about Kolmogorov complexity. We need more on this topic, brilliant episode!
Blaise is super interesting. I read his article called "Do large language models understand us?", and many of my assumptions were pleasantly challenged. This podcast was also a joy to listen to. Excited for the future.
I'm a regular listener on a podcast platform... but this episode was truly mindbending. I've forwarded links to it to several friends with "you really gotta listen to THIS one." I honestly think there's some kinda major intellectual revolution going on here, that is going to have repercussions and implications of first rank importance.
For me, the most exciting part of this work wasn't the creation of "life" from a random origin, but the creation of new complexity measures that are easy to compute and very sensitive to the presence of replicators. We've been in want of such measures for decades. This is a substantial step forward.
Wow. Somehow there were multiple points where I couldn’t tell if a point was trivial or profound. But what was profound was making it seem trivial. I truly understand how life is computation, and is ultimately inevitable in in the presence of energy and randomness. Wow. Truly a very set of wise words.
@@origins7298 Great point. Life, no, but we do see a self-perpetuating pattern/process (the nuclear fusion of the sun). I certainly wouldn't try to argue that the sun is alive without a few scooby snacks first, but its an interesting line of thought, which ultimately leads back to the question of "how do we even define what life is?" I think at least we would need natural selection.... What do you think, why don't we see any kind of selective forces occuring in the random and high energy environment of the sun?
That was both interesting and understandable, or at least understandable to the degree I can. In other words the explanations were clear and not filled with too much jargon.
Very thought-provoking discussion! I especially liked the illuminating description of the Turing machine. Somewhat reminiscent of Max Tegmark's "mathematical universe".
This episode was very interesting, please make more like these where you investigate how computer simulations are used to showcase Darwinian evolution.
Just fantastic this brilliant discussion between 2 very very clever scientists. This matter should be taught ( in a simplified version ) in high school to put humanity on the right track instead of sticking to century old dogmas.
Great podcast, but gonna go against the grain here. Initially, I found profound insights, but upon reflection, I see the research's value probably lies in its implications for computer science and AI development rather than for understanding life or origins. While the work is intriguing and Blaise’s software engineering expertise impressive, we should be cautious about drawing broad interdisciplinary conclusions. At its core, the research suggests that given enough random inputs, and effectively indefinite time, computer code sometimes replicates and changes, which I don’t find to be particularly surprising or enlightening about biology beyond general parallels. After all, art imitates life. I feel there’s far more work to be done before this research might lead to any accurate conclusions or insights about carbon based life and I’m frankly highly skeptical it will. Nonetheless, the discussion is thought-provoking and it’s interesting to know Google is working on these sorts of projects. Thanks for the content!
But doesn't computation happen only in a simulation? There's no computation happening in chemistry. It doesn't compute anything, it just happens. Copying in nature isn't computation, it's just how reality works. I feel that by inserting the term "computation" into the discussion we might be introducing an ultimately unhelpful analogy.
Makes me wish I was twenty years younger so I could see how this works out. I sense they are getting close and just need another 30 years to get there. Oh well. 😢
What does "a turing machine can compute anything anything else can compute" even mean? how do we know what even can be computed? don't we have to know the laws of physics for certain to answer that question? because so far, i've always thought of computable things to be defined as those that you can write a turing machine for, which leads to a circular definition.
## The Turing Machine and the Limits of Computation ### What does "a Turing machine can compute anything anything else can compute" mean? This statement is a bold claim, often referred to as the **Church-Turing thesis**. It essentially posits that a Turing machine, a theoretical computing device with a simple design, is capable of simulating any computation that any other computing device, no matter how complex, can perform. To put it simply, anything that can be calculated or solved by a computer, in theory, can also be calculated or solved by a Turing machine. ### How do we know what even can be computed? This is a profound question that has occupied philosophers and mathematicians for centuries. We don't have a definitive answer, but the Turing machine has provided a remarkably useful framework. The idea is this: * **Define computation:** We define computation as a process that follows a well-defined set of rules (an algorithm) to produce an output from a given input. * **Turing machine as a model:** We then define a Turing machine, a simple abstract model of computation. * **Hypothesis:** We hypothesize that any computation can be expressed as a Turing machine. This hypothesis has been incredibly robust. Despite decades of advancements in computer science, no one has found a computational problem that cannot be expressed as a Turing machine. This doesn't mean we've proven it definitively, but it's strong evidence. ### The Circularity Problem and the Laws of Physics You're right to point out the circularity in defining computable things as those that can be written as a Turing machine. It's a valid concern. However, the power of the Turing machine lies in its simplicity and universality. It provides a clear and consistent foundation for studying computability. As for the laws of physics, they certainly play a role in determining what is physically computable. For example, the speed of light imposes limits on how fast information can travel. However, the Turing machine is a theoretical concept, not a physical device. It operates in an idealized world without such constraints. **In conclusion,** while the Church-Turing thesis is a hypothesis, it's a remarkably resilient one. It has been a cornerstone of computer science and has helped us understand the limits of computation.
@@jjjccc728 all i know about is that it has been proven that if QM is true, then the church-turing thesis is true. but i often see people take that thesis for granted without ever mentioning it.
Mark Ruthy is tough , im lovin her views ! Youve always been 1 of my favorites but . Did u see the king of Jordan , "we will not b held hostage by Israel " thats exactly what Ruthy is talking about ! The King , pish's in his pants every time he has to man up against iran or iraq . How soon he forgets how Israel has bailed him out n propped him up ! Israel holding the region hostage ! Lot of nerve !
That's just a handcrafted demonstration of universal turing machines. This is different in that it just throws compute at the problem and gets something analagous to life. It would be closer to starting game of life completely randomly and getting game of life, which has never been done.
This is not ok. We are all intelligent enough to understand that the old adage/rhyme "Sticks and stones..." holds no real weight. To stress this argument, the normalization of such language increases the normalization of behavioral patterns. An example of this is, unfortunately, here in the UK we have young children (below the ages of 12 and some as young as 7 years of age), raping each other and recording it on Snapchat to send to their friends. This language needs to have a rethink or, more to the point, the creators need to understand the impact of their choices on the wider world at large. Sorry, I did try not to say anything, I just simply couldn't. They need a reality check, so, please, can you make sure they read this. Thank you 🙏
I knew Blaise when he was dropping out of the Princeton PhD physics program because he had invented foveated rendering and a wicked fast general purpose mipmapped UI. I’ve met some very bright people in my life but Blaise is at the very top of that list. I’m proud of him and glad to see his mind is running free at Google.
Excellent podcast. No way to walk away from this conversation without something to think about.
The most interesting talk I've heard in the last 20 years. So many things Blaise speaks about observing in his simulations have been speculated on in the origin of life community for decades (such as life-like autocatalytic chemistries being attractors in chemical space). We live in interesting times!
absolutely fantastic
What a great conversation - thanks a million for posting! I would love a discussion between Steven Wolfram and Blaise Agüera y Arcas moderated by Mr. Carroll.....
Great conversation between two great minds. One note on Kolmogorov complexity being easy to approximate: this is actually a concept most people get partially wrong since we can only really put an upper bound on the Kolmogorov complexity (say of a string, unless it’s a very simple one) so using techniques like Lempel-Ziv or Levin’s KT are really more of upper bounds. That being said I am guilty of that myself as I often use the word approximation when talking about Kolmogorov complexity. We need more on this topic, brilliant episode!
Blaise is super interesting. I read his article called "Do large language models understand us?", and many of my assumptions were pleasantly challenged. This podcast was also a joy to listen to. Excited for the future.
I'm a regular listener on a podcast platform... but this episode was truly mindbending. I've forwarded links to it to several friends with "you really gotta listen to THIS one." I honestly think there's some kinda major intellectual revolution going on here, that is going to have repercussions and implications of first rank importance.
For me, the most exciting part of this work wasn't the creation of "life" from a random origin, but the creation of new complexity measures that are easy to compute and very sensitive to the presence of replicators. We've been in want of such measures for decades. This is a substantial step forward.
Very interesting and thought provoking conversation! Thanks for sharing it!
Thanks for the fun discussion on a very interesting topic! Glad someone suggested this podcast to me
Wow. Somehow there were multiple points where I couldn’t tell if a point was trivial or profound. But what was profound was making it seem trivial. I truly understand how life is computation, and is ultimately inevitable in in the presence of energy and randomness. Wow. Truly a very set of wise words.
What about the sun? Plenty of energy and plenty of randomness there, no life.
@@origins7298 Great point. Life, no, but we do see a self-perpetuating pattern/process (the nuclear fusion of the sun). I certainly wouldn't try to argue that the sun is alive without a few scooby snacks first, but its an interesting line of thought, which ultimately leads back to the question of "how do we even define what life is?" I think at least we would need natural selection....
What do you think, why don't we see any kind of selective forces occuring in the random and high energy environment of the sun?
That was both interesting and understandable, or at least understandable to the degree I can. In other words the explanations were clear and not filled with too much jargon.
Very thought-provoking discussion! I especially liked the illuminating description of the Turing machine. Somewhat reminiscent of Max Tegmark's "mathematical universe".
Fantastic conversation about the most fascinating topic. Thank you.
Downloaded this paper a few days ago, but I haven't read it yet. This is a happy coincidence!
Sara Imari Walker mentioned this and therefore I am very intrigued. Thank you for this great podcast Sean and Blaise
Great ending. I have been trying to articulate this concept to people.
An especially enjoyable episode, thank you!
This episode was very interesting, please make more like these where you investigate how computer simulations are used to showcase Darwinian evolution.
Just fantastic this brilliant discussion between 2 very very clever scientists. This matter should be taught ( in a simplified version ) in high school to put humanity on the right track instead of sticking to century old dogmas.
I am surprised that neither of you mentioned Core Wars, which is a granddaddy of the virtual organism hosting sandboxes!
Your tips have helped me become a more successful trader and I got 13 out of 15 - a huge improvement for me. Thank you for your invaluable guidance
I remember the first time I did a double take on, "Fock Space." :D
First impression, wow his voice is so nice he could have been a voice actor
Outstanding
This was soo f*cking brain illuminating 😂❤
Great podcast, but gonna go against the grain here. Initially, I found profound insights, but upon reflection, I see the research's value probably lies in its implications for computer science and AI development rather than for understanding life or origins. While the work is intriguing and Blaise’s software engineering expertise impressive, we should be cautious about drawing broad interdisciplinary conclusions. At its core, the research suggests that given enough random inputs, and effectively indefinite time, computer code sometimes replicates and changes, which I don’t find to be particularly surprising or enlightening about biology beyond general parallels. After all, art imitates life. I feel there’s far more work to be done before this research might lead to any accurate conclusions or insights about carbon based life and I’m frankly highly skeptical it will. Nonetheless, the discussion is thought-provoking and it’s interesting to know Google is working on these sorts of projects. Thanks for the content!
But doesn't computation happen only in a simulation? There's no computation happening in chemistry. It doesn't compute anything, it just happens. Copying in nature isn't computation, it's just how reality works.
I feel that by inserting the term "computation" into the discussion we might be introducing an ultimately unhelpful analogy.
Wow, this episode was truly a brain fuck.
hey, I'll take some of that :)
Holy sound quality! Is this irl podcast?
Makes me wish I was twenty years younger so I could see how this works out. I sense they are getting close and just need another 30 years to get there. Oh well. 😢
I feel that the 'god of the gaps' is once again shrinking.
I wonder what they will call it once the gap disappears?? god of the wails...??
40:17 How to replicate
What does "a turing machine can compute anything anything else can compute" even mean? how do we know what even can be computed? don't we have to know the laws of physics for certain to answer that question? because so far, i've always thought of computable things to be defined as those that you can write a turing machine for, which leads to a circular definition.
en.wikipedia.org/wiki/Computable_number
## The Turing Machine and the Limits of Computation
### What does "a Turing machine can compute anything anything else can compute" mean?
This statement is a bold claim, often referred to as the **Church-Turing thesis**. It essentially posits that a Turing machine, a theoretical computing device with a simple design, is capable of simulating any computation that any other computing device, no matter how complex, can perform.
To put it simply, anything that can be calculated or solved by a computer, in theory, can also be calculated or solved by a Turing machine.
### How do we know what even can be computed?
This is a profound question that has occupied philosophers and mathematicians for centuries. We don't have a definitive answer, but the Turing machine has provided a remarkably useful framework.
The idea is this:
* **Define computation:** We define computation as a process that follows a well-defined set of rules (an algorithm) to produce an output from a given input.
* **Turing machine as a model:** We then define a Turing machine, a simple abstract model of computation.
* **Hypothesis:** We hypothesize that any computation can be expressed as a Turing machine.
This hypothesis has been incredibly robust. Despite decades of advancements in computer science, no one has found a computational problem that cannot be expressed as a Turing machine. This doesn't mean we've proven it definitively, but it's strong evidence.
### The Circularity Problem and the Laws of Physics
You're right to point out the circularity in defining computable things as those that can be written as a Turing machine. It's a valid concern. However, the power of the Turing machine lies in its simplicity and universality. It provides a clear and consistent foundation for studying computability.
As for the laws of physics, they certainly play a role in determining what is physically computable. For example, the speed of light imposes limits on how fast information can travel. However, the Turing machine is a theoretical concept, not a physical device. It operates in an idealized world without such constraints.
**In conclusion,** while the Church-Turing thesis is a hypothesis, it's a remarkably resilient one. It has been a cornerstone of computer science and has helped us understand the limits of computation.
@@jjjccc728 all i know about is that it has been proven that if QM is true, then the church-turing thesis is true. but i often see people take that thesis for granted without ever mentioning it.
@@garyl537 this wikipedia article doesn't answer my question
@@opensocietyenjoyerdo your own research until you will be able to understand turing completeness. Minecraft is Turing complete, for example.
YT's Closed Captions are calling it Brain Focus! 🤣🤣
{:o:O:}
"Brain focus"??
The stupid moralistic fucks!! for one of the most common words in the language. LOL
I lolled as soon as he said it :)
Mark Ruthy is tough , im lovin her views ! Youve always been 1 of my favorites but . Did u see the king of Jordan , "we will not b held hostage by Israel " thats exactly what Ruthy is talking about ! The King , pish's in his pants every time he has to man up against iran or iraq . How soon he forgets how Israel has bailed him out n propped him up ! Israel holding the region hostage ! Lot of nerve !
Oops wrong reply ,sowwee
Bad language! Rock n roll physicists smoking cigarettes and punching electrons.
Conways game of Life has been recreated in Conway's game of life already
That's just a handcrafted demonstration of universal turing machines. This is different in that it just throws compute at the problem and gets something analagous to life. It would be closer to starting game of life completely randomly and getting game of life, which has never been done.
This is not ok. We are all intelligent enough to understand that the old adage/rhyme "Sticks and stones..." holds no real weight.
To stress this argument, the normalization of such language increases the normalization of behavioral patterns.
An example of this is, unfortunately, here in the UK we have young children (below the ages of 12 and some as young as 7 years of age), raping each other and recording it on Snapchat to send to their friends.
This language needs to have a rethink or, more to the point, the creators need to understand the impact of their choices on the wider world at large.
Sorry, I did try not to say anything, I just simply couldn't.
They need a reality check, so, please, can you make sure they read this.
Thank you 🙏