YouSaid · the spoken record

Scott Aaronson

lines on the record: 134
first: 2020-10-12
most recent: 2020-10-12
sittings or episodes: 1
sources: podcast

Every line below is reproduced as it was said and linked to the record it came from. Nothing here is summarised or generated. Directory · Search · Corrections

“Let me tell you sort of the biggest ones, the ones that you would learn first. So first of all, there is P. That's what it's called. It stands for polynomial time. And this is just the class of all of the problems that you could solve with a conventional computer like your iPhone or your laptop, you know, by a completely deterministic algorithm, right? Using a number of steps that grows only like the size of the input raised to some fixed power.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Well, you could, but beyond a certain point, they become unstable, right? Right. So it's like, you know, in theory, you can have atoms with, you know, and look, look, I mean, a neutron star, you know, is a nucleus with, you know, untouched billions of neutrons in it, of hadrons in it. Okay, but for sort of normal atoms, right, probably you can't get much above 100 atomic weight 150 or so, or sorry, sorry, I mean, I mean, beyond 150 or so protons without it, you know, very quickly fissioning. With complexity classes, well, yeah, you can have an infinity of complexity classes. But, you know, maybe.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Right. So complexity theory is a huge part of, let's say, the theoretical core of computer science. It started in the 60s and 70s as sort of an autonomous field. So it was already, you know, I mean, it was well developed even by the time that I was born. In 2002, I made a website called the Complexity Zoo to answer your question, where I just tried to catalog the different complexity classes, which are classes of problems that are solvable with different kinds of resources. Okay. So these are kind of, you know, you could think of complexity classes as like being almost too theoretical computer science, like what the elements are to chemistry, right? They're sort of, you know, there are our most basic objects. In a certain way”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Okay, but if that's true, then in some ways that's reassuring because if that's the best that they can do, then that would say that they can't break 2,000 bit numbers.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, exactly. So the fastest algorithms that anyone has discovered, at least publicly discovered, you know, I'm assuming that the NSA doesn't know something better. Okay, but they take time that basically grows exponentially with the cube root of the size of the number that you're factoring, right? So that cube root, that's the part that takes all the cleverness. Okay, but there's still an exponentiality there. What that means is that when people use a thousand bit keys for their cryptography, that can probably be broken using the resources of the NSA or the world's other intelligence agencies. You know, people have done analyses that say, you know, with a few hundred million dollars of computer power, they could totally do this. And if you look at the documents that Snowden released, it looks a lot like they are doing that or something like that. It would kind of be surprising if they weren't.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Number of factors. Now, and as many people might know, for better or worse, the security of most of the encryption that we currently use to protect the internet is based on the belief, and this is not a theorem, it's a belief, that factoring is an inherently hard problem for our computers. We do know algorithms that are better than just trial division and just trying all the possible divisors, but they are still basically exponential.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Now, as it turns out, people discovered much faster ways to multiply numbers using computers. And today we know how to multiply two numbers that are n digits long using a number of steps that's nearly linear in n. These are questions you can ask. But now let's think about a different thing that people have encountered in elementary school factoring a number. Take a number and find its prime factors. And here, you know, if I give you a number with 10 digits, I ask you for its prime factors. Well, maybe it's even so you know that 2 is a factor. Maybe it ends in zero. So you know that 10 is a factor, right? But, you know, other than a few obvious things like that, if the prime factors are all very large, then it's not clear how you even get started, right? You know, it seems like you have to do an exhaustive search among an enormous.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, that's right. If you just use the elementary school algorithm of just carrying, you know, then it takes time that is linear in the length of the numbers, right? Now, multiplication, if you use the elementary school algorithm, is harder because you have to multiply each digit of the first number by each digit of the second one and then deal with all the carries. So that's what we call a quadratic time algorithm, right? If the numbers become twice as long, now you need four times as much time.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Oh, alright, alright, alright, alright. Complexity theory is the study of sort of the inherent resources needed to solve computational problems. It's easiest to give an example. Like, let's say we want to add two numbers, right? If I want to add them, if the numbers are twice as long, then it will take me twice as long to add them, but only twice as long, right? It's no worse than that.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Well, yeah, there's the question of how much time of how hard is it to write a program. And then there's also the questions of what resources does the program need? How much time, how much memory? Those are much more complicated questions. Of course, ones that we're still struggling with today.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Right, and you know, and then how would you do that? Obviously, you couldn't do it itself in basic, right? But there is this incredible flattening that happens once you learn what is universality, but then it's also like an opportunity because it means once you know these rules, then the sky is the limit, right? Then you have kind of the same weapons at your disposal that the world's greatest programmer has. It's now all just a question of how you wield them.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Well, I mean, I mean, right. I mean, in one sense, it was this. I had imagined that there was going to be some infinite hierarchy of more and more powerful programming languages. And then I kicked myself for having such a stupid idea. But apparently Girdle had had the same conjecture in the 30s. Oh, good. You're in good company. And then Girdle read Touring's paper and he kicked himself and he said, yeah, I was completely wrong about that. But, you know, I had thought that maybe where I can contribute will be to invent a new, more powerful programming language that lets you express things that could never be expressed in basic.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Okay, and he but he proved that this could simulate all kinds of other things. Well, we would call it a Turing universal model of computation that is, you know, just as it has just the same expressive power that basic or Java or C++ or any of those other languages have because anything in those other languages could be compiled down to Turing machine. Now Turing also proved a different related thing which is that there is a single Turing machine that can simulate any other Turing machine if you just describe that other machine on its tape and likewise there is a single Turing machine that will run any C program if you just put it on its tape. That's a second meaning of universality.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“So this idea of universality goes back, at least to Alan Turing in the 1930s, when he wrote down this incredibly simple pared-down model of a computer, the Turing machine, right? Which, you know, he pared down the instruction set to just read a symbol, you know, write a symbol, move to the left, move to the right, halt, change your internal state. Right, that's it.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Can build the universe out of Nand Gates. Yeah. You know, the simple instructions of BASIC are already enough, at least in principle, if we ignore details like how much memory can be accessed and stuff like that. That is enough to express what could be expressed by any programming language whatsoever. And the way to prove that is very simple. We simply need to show that in basic or whatever, we could write an interpreter or a compiler for whatever other programming language we care about, like C or Java or whatever. And as soon as we had done that, then Ipso facto, anything that's expressible in C or Java is also expressible in BASIC.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Know it's like finding out where babies come from. It's like that level of, you know, why didn't anyone tell me this before, right? But I thought, okay, this is just the beginning. Now I know how to write a basic program. But, you know, really write an interesting program, like, you know, a video game, which had always been my dream as a kid, create my own Nintendo games, right? But obviously I'm going to need to learn some way more complicated form of programming than that. You know, eventually I learned this incredible idea of universality. And that says that no, you throw in a few rules and then you can, you already have enough to express everything. Okay, so for example, the and, the or, and the not gate can all, or in fact, even just the and the not gate, or even just the nand gate, for example, is already enough to express any Boolean function on any.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“General have I think I'm gonna have to go with the idea of universality. You know, if you're really asking for the most beautiful, I mean, so universality is the idea that, you know, you put together a few simple operations. In the case of Boolean logic, that might be the end gate, the or gate, the not gate, right? And then your first guess is, okay, this is a good start. But obviously, as I want to do more complicated things, I'm going to need more complicated building blocks to express that, right? And that was actually my guess when I first learned what programming was. I mean, when I was, you know, an adolescent and someone showed me Apple Basic. And GW Basic, if anyone listening remembers that. Okay, but, you know, I thought, okay, well, now, you know, I mean, I thought I felt like this is a revelation.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Oh, yeah, no, absolutely. Although, like, if we ask it to a word problem that involves reasoning about the locations of things in space, I don't think it does such a great job on those, right? To take an example. And so the guess would be, well, humans have a lot of predictive processing, a lot of just filling in the blanks. But we also have these other mechanisms that we can couple to or that we can sort of call as subroutines when we need to. And that maybe, you know, to go further, that one would want to integrate other forms of reasoning.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Should stop doing that and start doing something very different, like some more exact logical reasoning, right? And so, you know, one is naturally led to guess that our brain sort of has some element of predictive processing, but that it's coupled to other mechanisms, right? That it's coupled to, you know, first of all, visual reasoning, which GPT-3 also doesn't have any of, right? Although there's some”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, I mean, some people say. Yes, yeah, so sure. So there's certainly that limit. I mean, there's also, you know, like if you are looking for questions that will stump GPT-3, right, you can come up with some without even getting it to learn how to balance parentheses. It's failures are ironic, right? Like basic arithmetic, right? Arithmetic. And you think, you know, isn't that what computers are supposed to be best at? Isn't that where computers already had us beat a century ago? Right. And yet that's where GPT-3 struggles, right? But it's amazing, you know, that it's almost like a young child in that way, right? But somehow, you know, because it is just trying to predict what comes next, it doesn't know when it's...”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“I mean, the truth is that we don't really know what the limits are, right? Exactly. Because, you know, what we've seen so far is that GPT-3 was basically the same thing as GPT-2, but just with a much larger network, more training time, bigger training corpus. And it was very noticeably better than its immediate predecessor. So we don't know where you hit the ceiling here, right? I mean, that's the amazing part and maybe also the scary part, right? That.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“But yeah, but I mean, I mean, and sometimes that is, you know, if there is a deep philosophical question that's raised by GPT-3, then that is it, right? Are we doing anything other than, you know? This predictive processing just trying to constantly trying to fill in a blank of what would come next after what we just said up to this point. Is that what I'm doing right now?”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“And all that kind of stuff, but basically, you know, it is a neural net, you know. So, I mean, I mean, or what's now called a deep net, but, you know, they're basically the same thing, right? So it's a form of, you know, algorithm that people have known about for decades, right? But it is constantly trying to solve the problem, predict the next word, right? So it's just trying to predict what comes next. It's not trying to decide what it should say, what ought to be true. It's trying to predict what someone who had said all of the words up to the preceding one would say next.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Well, okay. I mean, the training involved massive amounts of data from the internet and actually took lots and lots of computer power, lots of electricity, right? You know, there are some very prosaic reasons why this wasn't done earlier, right? But it costs some tens of millions of dollars, I think, just for approximately.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“A chatbot by just doing deep learning over a corpus consisting of the entire web Right, and so now they finally have done that, right? And, you know, the results are very impressive. You know, it's not clear that, you know, people can argue about whether this is truly a step toward general AI or not, but this is an amazing capability that we didn't have a few years ago that a few years ago, if you had told me that we would have it now, that would have surprised me. And I think that anyone who denies that is just not engaging with what's there.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“And it will do, you know, not a perfect, not a great job, not an amazing job, but a passable job, you know, definitely, you know, as good as, you know, in many cases, I would say better than I would have done, right? You know, you can ask it to write an essay, like a student essay about pretty much any topic, and it will get something that I am pretty sure would get at least a B minus, you know, in the most, you know, high school or even college classes, right? And, you know, in some sense, you know, the way that it did this, the way that it achieves this, you know, Scott Alexander of the much more blog, Slate Star Codex, had a wonderful way of putting it. He said that they basically just ground up the entire internet into a slurry. And, you know, and to tell you the truth, I had wondered for a while why nobody had tried that, right? Like, why not run?”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, and the last few months, we've had, you know, the world has now seen a chat engine or a text engine, I should say, called GPT-3. I think it's still, you know, it does not pass a Turing test. You know, there are no real claims that it passes the Turing test, right? You know, this comes out of the group at OpenAI. And, you know, they've been relatively careful in what they've claimed about the system. But I think this. As clearly as Eugene Gustman was not in advance over Eliza, it is equally clear that this is a major advance over Eliza or really over anything that the world has seen before. This is a text engine that can Come up with kind of on topic, reasonable sounding completions to just about anything that you ask. You can ask it to write a poem about topic X in the style of poet Y, and it will have a go at that.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“You know, you shouldn't ask me. Then they said, Look, you can go here and you can try it out. I said, All right, all right. So I'll try it out. But now, you know, this whole discussion, I mean, it got a whole lot more interesting in just the last few months.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“You know, it really had some people convinced. But, you know, this thing was just like, I think it was literally just a few hundred lines of Lisp code, right? It was not only was it not intelligent, it wasn't especially sophisticated. It was like a simple little hobbyist program. And Eugene Gustman, from what I could see, was not a significant advance compared to Eliza, right? And that was really the point I was making. And this was, you know, you didn't, in some sense, you didn't need a like a computer science professor to sort of say this, like anyone who was looking at it and who just had, you know, an ounce of sense could have said the same thing, right? But because, you know, these journalists were calling me, you know, like the first thing I said was, well, you know, no, you know, I'm a quantum computing person. I'm not an AI person.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Right, right. It was not able to answer or even intelligently respond to basic common sense questions. But let me say something stronger than that. There was a famous chatbot in the 60s called Eliza, right? That, you know, that managed to actually fool.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“I had a sort of different way to look at it instead of the legalistic way. Let's just try the actual thing out and let's see what it can do with questions like, you know, is Mount Everest bigger than a shoebox. Or just, you know, like the most obvious questions, right? And then, and, you know, and the answer is, well, it just kind of parries you because it doesn't know what you're talking about, right?”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“So, yeah. So, all that happened was that a bunch of journalists started writing breathless articles about the first chatbot that passes the Turing test, right? And it was this thing called Eugene Gustman that was supposed to simulate a 13-year-old boy. And apparently someone had done some test where, you know, people couldn't were less than perfect, let's say, distinguishing it from a human. And they said, well, if you look at Turing's paper and you look at, you know, the percentages that he talked about, then it seemed like we're past that threshold, right?”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Not really. I didn't. I mean, there was this kind of ridiculous claim that was made some almost a decade ago about a chat bot called Eugene Goose budget.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“It is perfectly plausible to imagine a computer that could say it would not be limited to working within a single formal system, right? They could say, I am now going to adopt the hypothesis that my formal system is consistent. And I'm now going to see what can be done from that stronger vantage point and so on. And yeah, when I'm going to add new axioms to my system. Totally plausible. There's absolutely Girdle's theorem has nothing to say about against an AI that could repeatedly add new axioms. All it says is that there is no... Absolute guarantee that when the AI adds new axioms that it will always be right And that's, of course, the point that Penrose pounces on. But the reply is obvious. And it's one that Alan Toring made 70 years ago. Namely, we don't have an absolute guarantee that we're right when we add a new axiom.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Okay, but the basic problem with this idea is, you know, Penrose wants to say that and all of his predecessors here, you know, want to say that, you know, even though this given formal system cannot prove its own consistency, we as humans are sort of looking at it from the outside can just somehow see its consistency, right? And the rejoinder to that, you know, from the very beginning has been, well, can we really? I mean, maybe he, Penrose can, but can the rest of us, right? And, you know, I noticed that”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Right, no good formal system can actually prove its own consistency. That can only be done from a stronger formal system, which then can't prove its own consistency and so on forever. Okay, that Girdle's theorem. But now why is that relevant to consciousness? Well, you know, I mean, the idea that it might have something to do with consciousness as an old one, Girdle himself apparently thought that it needed. You know, Lucas thought so, I think, in the 60s. And Penrose is really just, you know, sort of updating what they and others had said. I mean, you know, the idea that Girdle's theorem could have something to do with consciousness was, you know, in 1950 when Alan Turing wrote his article about the Turing test, he already was writing about that as an old and well-known idea.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Is that where it takes me? No, it's not just that because, I mean, everyone agrees that there are problems that are uncomputable, right? That's a mathematical theorem, right? But what Penrose wants to say is that, for example, there are statements, given any formal system, you know, for doing math, there will be true statements of arithmetic that that formal system, you know, if it's adequate for math at all, if it's consistent and so on, will not be able to prove a famous example being the statement that that system itself is consistent.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Spontaneously collect. For consciousness to somehow be able to influence the direction of the collapse so that it wouldn't be completely random, but that your dispositions would somehow influence the quantum state to collapse more likely this way or that way. Okay Finally, Penrose says that all of this has to be true because of an argument that he makes based on girdle's incompleteness theorem. Now, Mike, I would say the overwhelming majority of computer scientists and mathematicians who have thought about this, I don't think that girdles incompleteness theorem can do what he needs it to do here, right? I don't think that that argument is sound. But that is sort of the tower that you have to ascend to if you're going to go where Penrose goes. And the intuition users.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Totally adequate for all of the terrestrial phenomena, right? The only things that it doesn't, you know, explain are, well, first of all, you know, the details of gravity, if you were to probit, like at extremes of curvature or incredibly small distances, it doesn't explain dark matter. It doesn't explain black hole singularities, right? But these are all very exotic things, very, you know, far removed from our life on Earth. So for Penrose to be right, he needs these phenomena to somehow affect the brain. He needs the brain to contain antenna that are sensitive to the black holes, to this as yet unknown physics, right? And then he needs a modification of quantum mechanics. So he needs quantum mechanics to actually be wrong. He needs what he wants is what he calls an objective.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Effects are coming from quantum gravity, you know, they are sort of screened off by quantum field theory, right? And this brings us to the whole idea of effective theories, right? But we have in the standard model of elementary particles, right? We have a quantum field theory that seems”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“But, all right, but supposing that he's right about that, then what most physicists would say is that whatever new phenomena there are in quantum gravity, they might be relevant at the singularities of black holes. They might be relevant at the Big Bang. They are plainly not relevant to something like the brain that is operating at ordinary temperatures with ordinary chemistry and the physics underlying the brain, they would say that we have, you know, the fundamental physics of the brain, they would say that we've pretty much completely known for generations now, right? Because quantum field theory lets us sort of parameterize our ignorance. Sean Carroll has made this case and, you know, in great detail, right? That sort of whatever new”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Okay, yes, yes. That would be literally uncomputable. And I've asked him, you know, to clarify this, but uncomputable, even if you had an oracle for the halting problem or as high up as you want to go in the sort of the usual hierarchy of uncomputability. He wants to go beyond all of that. Okay, so just to be clear, if we're keeping count of how many speculations, there's probably like at least five or six of them, right? There's first of all that there is some quantum gravity theory that would involve this kind of uncomputability, right? Most people who study quantum gravity would not agree with that. They would say that what we've learned, you know, what little we know about quantum gravity from this ADS-CFT correspondence, for example, has been very much consistent with the broad idea of nature being computable, right?”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“You know, in fact, a quantum computer can be simulated by an ordinary computer. It might merely need exponentially more time in order to do so, right? So that's simply not good enough for him. Okay, so what he wants is for the brain to be a quantum gravitational computer. Or he wants the brain to be exploiting as yet unknown laws of quantum gravity, which would be uncomputable.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Of course. You know, I read Penrose's books as a teenager. They had a huge impact on me. Five or six years ago, I had the privilege to actually talk these things over with Penrose. At some length at a conference in Minnesota. And, you know, he is an amazing personality. I admire the fact that he was even raising such audacious questions at all. But to answer your question, I think the first thing we need to get clear on is that he is not merely saying that quantum mechanics is relevant to consciousness, right? That would be like tame compared to what he is saying, right? He is saying that even quantum mechanics is not good enough, right? Because if supposing, for example, that the brain were a quantum computer, that's still a computer.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Apart from you, right? It says the same kinds of things that you would have said, maybe not exactly the same things because we agree that there's noise, but it says the same kinds of things. And maybe you alone would say, no, I know that that's not me. You know, it doesn't share my, I haven't felt my consciousness leap over to that other thing. I still feel it localized in this version, right? Then why should anyone else believe you?”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, yeah, yeah. And especially once we say that, well, look, maybe there's no way to make a deterministic prediction because we know that there's noise buffeting the brain around, presumably even quantum mechanical uncertainty affecting the sodium ion channels, for example, whether they open or they close. There's no reason why over a certain time scale that shouldn't be amplified, just like we imagine happens with the weather or with any other, you know, chaotic system. So if that stuff is important, right, then we would say, well, you can't, you know, you're never going to be able to make an accurate enough copy. But now the hard part is, well, what if someone can make a copy that sort of no one else can tell?”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Well, I don't know. I mean, it might be possible, but only at the cost of destroying the person. I mean, it depends on how low you have to go in sort of the substrate. Like if there was a nice digital abstraction layer, if you could think of each neuron as a kind of transistor computing a digital function, then you could imagine some nanorobots that would go in and would just scan the state of each transistor, you know, of each neuron. And then make a good enough copy, right? But if it was actually important to get down to the molecular or the atomic level, then eventually you would be up against quantum effects. You would be up against the unclonability of quantum states. So I think it's a question of how good of a good does the replica have to be before you're going to count it as actually a copy of you or as being able to predict your actions.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Conversation could possibly do that. And so I think it becomes extremely interesting to ask could such predictions be done, you know, even in principle, is it consistent with”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“Let me put it this way. If you could do, like, in a Greek tragedy where you would just write down a prediction for what I'm going to do and then maybe you put the prediction in a sealed box and maybe, you know, you open it later and you show that you knew everything I was going to do. Or, you know, of course, the even creepier version would be you tell me the prediction and then I try to falsify it and my very effort to falsify it makes it come true. Let's even forget that version as convenient as it is for fiction writers. Let's just do the version where you put the prediction into a sealed envelope. If you could reliably predict everything that I was going to do, I'm not sure that that would destroy my sense of being conscious, but I think it really would destroy my sense of having free will, you know, and much, much more than any philosophical”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source
“A property of the brain, if true, that we distinguish it in a principled way, at least from any currently existing computer, not from any possible computer, but from, yeah, yeah.”
2020-10-12 · Lex Fridman Podcast · #130 – Scott Aaronson: Computational Complexity and Consciousness · IDENTIFIED FROM THE TRANSCRIPT · source