Scott Aaronson: Computational Complexity and Consciousness

00:00:00.000 | The following is a conversation with Scott Aronson, his second time on the podcast.

00:00:04.240 | He is a professor at UT Austin, director of the Quantum Information Center,

00:00:10.320 | and previously a professor at MIT. Last time we talked about quantum computing,

00:00:15.920 | this time we talk about computational complexity, consciousness, and theories of everything.

00:00:24.160 | I'm recording this intro, as you may be able to tell, in a very strange room in the middle of the

00:00:32.240 | night. I'm not really sure how I got here or how I'm going to get out, but Hunter S. Thompson saying,

00:00:40.560 | I think, applies to today and the last few days and actually the last couple of weeks.

00:00:46.880 | "Life should not be a journey to the grave with the intention of arriving safely in a pretty and

00:00:53.680 | well-preserved body, but rather to skid in broadside in a cloud of smoke, thoroughly used up,

00:01:00.640 | totally worn out, and loudly proclaiming, 'Wow, what a ride.'" So I figured whatever I'm up to here,

00:01:10.400 | and yes, lots of wine is involved, I'm gonna have to improvise, hence this recording. Okay,

00:01:18.480 | quick mention of each sponsor, followed by some thoughts related to the episode.

00:01:23.200 | First sponsor is SimpliSafe, a home security company I use to monitor and protect my apartment,

00:01:28.720 | though of course I'm always prepared with a fallback plan, as a man in this world must always

00:01:36.720 | be. Second sponsor is 8Sleep, a mattress that cools itself, measures heart rate variability,

00:01:46.080 | has an app, and has given me yet another reason to look forward to sleep, including the all

00:01:52.880 | important PowerNap. Third sponsor is ExpressVPN, the VPN I've used for many years to protect my

00:01:59.840 | privacy on the internet. Finally, the fourth sponsor is BetterHelp, online therapy when you

00:02:07.200 | want to face your demons with a licensed professional, not just by doing David Goggins-like

00:02:12.800 | physical challenges like I seem to do on occasion. Please check out these sponsors in the description

00:02:18.800 | to get a discount and to support the podcast. As a side note, let me say that this is the second

00:02:25.280 | time I recorded a conversation outdoors. The first one was with Stephen Wolfram when it was actually

00:02:31.680 | sunny out, in this case it was raining, which is why I found a covered outdoor patio. But I learned

00:02:38.960 | a valuable lesson, which is that raindrops can be quite loud on the hard metal surface of a patio

00:02:45.840 | cover. I did my best with the audio, I hope it still sounds okay to you. I'm learning, always

00:02:52.320 | improving. In fact, as Scott says, if you always win, then you're probably doing something wrong.

00:02:59.120 | To be honest, I get pretty upset with myself when I fail, small or big, but I've learned that this

00:03:06.880 | feeling is priceless. It can be fuel when channeled into concrete plans of how to improve.

00:03:14.480 | So if you enjoy this thing, subscribe on YouTube, review the 5 Stars on Apple Podcasts,

00:03:20.320 | follow on Spotify, support on Patreon, or connect with me on Twitter @LexFriedman.

00:03:25.920 | And now, here's my conversation with Scott Aaronson.

00:03:29.760 | Let's start with the most absurd question, but I've read you write some fascinating stuff about

00:03:35.520 | it, so let's go there. Are we living in a simulation? What difference does it make,

00:03:40.800 | Lex? I mean, I'm serious. What difference?

00:03:43.760 | Because if we are living in a simulation, it raises the question, how real does something

00:03:49.360 | have to be in simulation for it to be sufficiently immersive for us humans?

00:03:53.920 | But I mean, even in principle, how could we ever know if we were in one, right? A perfect simulation,

00:03:59.920 | by definition, is something that's indistinguishable from the real thing.

00:04:03.200 | But we didn't say anything about perfect. It could be imperfect.

00:04:05.040 | No, no, that's right. Well, if it was an imperfect simulation, if we could hack it,

00:04:10.240 | you know, find a bug in it, then that would be one thing, right? If this was like the Matrix,

00:04:15.360 | and there was a way for me to, you know, do flying Kung Fu moves or something by

00:04:20.400 | hacking the simulation, well, then, you know, we would have to cross that bridge when we came to

00:04:24.400 | it, wouldn't we? Right? I mean, at that point, you know, it's hard to see the difference between that

00:04:31.360 | and just what people would ordinarily refer to as a world with miracles, you know?

00:04:36.800 | What about from a different perspective, thinking about the universe as a computation,

00:04:41.840 | like a program running on a computer? That's kind of a neighboring concept.

00:04:45.600 | It is. It is an interesting and reasonably well-defined question to ask,

00:04:50.480 | is the world computable? You know, does the world satisfy what we would call in CS the

00:04:56.880 | Church-Turing thesis? That is, you know, could we take any physical system and simulate it to,

00:05:04.640 | you know, any desired precision by a Turing machine, you know, given the appropriate input

00:05:09.680 | data, right? And so far, I think the indications are pretty strong that our world does seem to

00:05:16.160 | satisfy the Church-Turing thesis. At least if it doesn't, then we haven't yet discovered why not.

00:05:22.320 | But now, does that mean that our universe is a simulation? Well, you know, that word seems to

00:05:29.760 | suggest that there is some other larger universe in which it is running, right? And the problem

00:05:35.680 | there is that if the simulation is perfect, then we're never going to be able to get any direct

00:05:41.600 | evidence about that other universe. You know, we will only be able to see the effects of the

00:05:47.760 | computation that is running in this universe. Well, let's imagine an analogy. Let's imagine a

00:05:54.240 | PC, a personal computer, a computer. Is it possible with the advent of artificial intelligence for the

00:06:02.080 | computer to look outside of itself to see, to understand its creator? I mean, is that a

00:06:10.400 | ridiculous analogy? Well, I mean, with the computers that we actually have, I mean, first of all,

00:06:17.840 | we all know that humans have done an imperfect job of enforcing the abstraction boundaries of

00:06:25.280 | computers, right? You may try to confine some program to a playpen, but as soon as there's one

00:06:32.560 | memory allocation error in the C program, then the program has gotten out of that playpen and

00:06:39.920 | it can do whatever it wants, right? This is how most hacks work, you know, with the viruses and

00:06:45.600 | worms and exploits. And you would have to imagine that an AI would be able to discover something

00:06:51.840 | like that. Now, of course, if we could actually discover some exploit of reality itself,

00:06:58.640 | then in some sense, we wouldn't have to philosophize about this, right? This would

00:07:06.880 | no longer be a metaphysical conversation. This would just be a...

00:07:12.160 | But the question is, what would that hack look like?

00:07:15.120 | Yeah, well, I have no idea. I mean, Peter Shore, you know, a very famous person in quantum

00:07:22.640 | computing, of course, has joked that maybe the reason why we haven't yet integrated general

00:07:29.280 | relativity and quantum mechanics is that the part of the universe that depends on both of them was

00:07:35.520 | actually left unspecified. And if we ever tried to do an experiment involving the singularity of a

00:07:42.320 | black hole or something like that, then the universe would just generate an overflow error

00:07:47.680 | or something, right? A blue screen of death.

00:07:49.680 | Yeah, we would just crash the universe. Now, the universe has seemed to hold up pretty well for

00:07:58.080 | 14 billion years, right? So, you know, my Occam's razor kind of guess has to be that it will continue

00:08:08.560 | to hold up. That the fact that we don't know the laws of physics governing some phenomenon

00:08:14.480 | is not a strong sign that probing that phenomenon is going to crash the universe,

00:08:19.360 | right? But, you know, of course, I could be wrong.

00:08:22.720 | But do you think on the physics side of things, you know, there's been recently a few folks,

00:08:29.280 | Eric Weinstein and Stephen Wolfram, that came out with the theory of everything. I think there's a

00:08:33.920 | history of physicists dreaming and working on the unification of all the laws of physics. Do you

00:08:40.960 | think it's possible that once we understand more physics, not necessarily the unification of the

00:08:47.200 | laws, but just understand physics more deeply at the fundamental level, we'll be able to start

00:08:52.960 | you know, I mean, part of this is humorous, but looking to see if there's any bugs in the universe

00:08:59.920 | that could be exploited for, you know, traveling at not just speed of light, but just traveling

00:09:06.560 | faster than our current spaceships can travel, all that kind of stuff?

00:09:10.960 | Well, I mean, to travel faster than our current spaceships could travel, you wouldn't need to

00:09:15.920 | find any bug in the universe, right? The known laws of physics, you know, let us go much faster,

00:09:21.360 | up to the speed of light, right? And you know, when people want to go faster than the speed of

00:09:26.160 | light, well, we actually know something about what that would entail, namely that, you know,

00:09:31.200 | according to relativity, that seems to entail communication backwards in time, okay? So then

00:09:37.200 | you have to worry about closed time-like curves and all of that stuff. So, you know, in some sense,

00:09:42.000 | we sort of know the price that you have to pay for these things, right?

00:09:46.400 | But under the current understanding of physics.

00:09:48.880 | That's right, that's right. We can't, you know, say that they're impossible, but we know that

00:09:54.480 | sort of a lot else in physics breaks, right? So now, regarding Eric Weinstein and Stephen Wolfram,

00:10:02.960 | like I wouldn't say that either of them has a theory of everything. I would say that they have

00:10:07.920 | ideas that they hope, you know, could someday lead to a theory of everything.

00:10:12.400 | Is that a worthy pursuit?

00:10:13.760 | Well, I mean, certainly, let's say by theory of everything, you know, we don't literally mean a

00:10:19.200 | theory of cats and of baseball and, you know, but we just mean it in the more limited sense of

00:10:25.440 | everything, a fundamental theory of physics, right? Of all of the fundamental interactions

00:10:32.080 | of physics. Of course, such a theory, even after we had it, you know, would leave the entire

00:10:38.960 | question of all the emergent behavior, right? You know, to be explored. So it's only everything for

00:10:46.320 | a specific definition of everything. Okay, but in that sense, I would say, of course,

00:10:50.400 | that's worth pursuing. I mean, that is the entire program of fundamental physics,

00:10:55.360 | right? All of my friends who do quantum gravity, who do string theory, who do anything like that,

00:11:00.960 | that is what's motivating them.

00:11:02.800 | Yeah, it's funny though, but I mean, Eric Weinstein talks about this. It is,

00:11:07.120 | I don't know much about the physics world, but I know about the AI world. It is a little bit taboo

00:11:12.960 | to talk about AGI, for example, on the AI side. So really, to talk about the big dream of the

00:11:23.520 | community, I would say, because it seems so far away, it's almost taboo to bring it up because,

00:11:28.560 | you know, it's seen as the kind of people that dream about creating a truly superhuman level

00:11:34.400 | intelligence that's really far out there. People, because we're not even close to that.

00:11:39.200 | And it feels like the same thing is true for the physics community.

00:11:42.480 | I mean, Stephen Hawking certainly talked constantly about theory of everything, right?

00:11:48.880 | You know, I mean, people, you know, use those terms who were, you know, some of the most

00:11:54.240 | respected people in the whole world of physics, right? But I mean, I think that the distinction

00:12:00.880 | that I would make is that people might react badly if you use the term in a way that suggests

00:12:07.200 | that you, you know, thinking about it for five minutes have come up with this major new insight

00:12:12.880 | about it, right? It's difficult. Stephen Hawking is not a great example because I think you can do

00:12:21.200 | whatever the heck you want when you get to that level. And I certainly see like senior faculty,

00:12:28.720 | you know, at that point, that's one of the nice things about getting older is you stop giving a

00:12:34.160 | damn. But community as a whole, they tend to roll their eyes very quickly at stuff that's outside

00:12:40.080 | the quote unquote mainstream. Well, let me put it this way. I mean, if you asked, you know, Ed

00:12:44.960 | Witten, let's say, who is, you know, you might consider a leader of the string community and

00:12:49.760 | thus, you know, very, very mainstream in a certain sense, but he would have no hesitation in saying,

00:12:55.440 | you know, of course, you know, they're looking for a, you know, a unified description of nature,

00:13:03.120 | of, you know, of general relativity, of quantum mechanics, of all the fundamental interactions

00:13:08.480 | of nature, right? Now, you know, whether people would call that a theory of everything, whether

00:13:14.400 | they would use that term, that might vary. You know, Lenny Suskin would definitely have no

00:13:19.360 | problem telling you that, you know, if that's what we want, right? For me, who loves human beings

00:13:24.320 | and psychology, it's kind of ridiculous to say a theory that unifies the laws of physics gets you

00:13:32.720 | to understand everything. I would say you're not even close to understanding everything.

00:13:36.800 | Yeah, right. Well, yeah, I mean, the word everything is a little ambiguous here,

00:13:41.280 | right? Because, you know, and then people will get into debates about, you know, reductionism

00:13:46.080 | versus emergentism and blah, blah, blah. And so in not wanting to say theory of everything,

00:13:52.400 | people might just be trying to short circuit that debate and say, you know, look, you know,

00:13:57.040 | yes, we want a fundamental theory of, you know, the particles and interactions of nature.

00:14:02.480 | Let me bring up the next topic that people don't want to mention, although they're getting more

00:14:05.920 | comfortable with it, is consciousness. You mentioned that you have a talk on consciousness

00:14:10.240 | that I watched five minutes of, but the internet connection is really bad.

00:14:14.160 | Was this my talk about, you know, refuting the integrated information theory?

00:14:18.640 | Yes, it might have been.

00:14:18.800 | Which was this particular account of consciousness that, yeah, I think

00:14:21.840 | one can just show it doesn't work. But it's much harder to say what does work.

00:14:25.600 | What does work, yeah. Let me ask, maybe it'd be nice to comment on,

00:14:30.160 | you talk about also like the semi-hard problem of consciousness, or like almost hard,

00:14:35.760 | probably, or kind of hard.

00:14:36.800 | Pretty hard.

00:14:37.600 | Pretty hard. So maybe can you talk about that, their idea of the approach to modeling consciousness

00:14:47.040 | and why you don't find it convincing? What is it, first of all?

00:14:49.680 | Okay, well, so what I call the pretty hard problem of consciousness, this is my term,

00:14:55.920 | although many other people have said something equivalent to this, okay? But it's just, you know,

00:15:02.400 | the problem of, you know, giving an account of just which physical systems are conscious and

00:15:09.840 | which are not. Or, you know, if there are degrees of consciousness, then quantifying how conscious

00:15:15.840 | a given system is.

00:15:17.280 | Oh, awesome. So that's the pretty hard problem.

00:15:19.200 | Yeah, that's what I mean by it.

00:15:20.400 | That's it, I'm adopting it. I love it. That's a good ring to it.

00:15:23.680 | And so, you know, the infamous hard problem of consciousness is to explain how something like

00:15:29.760 | consciousness could arise at all, you know, in a material universe, right? Or, you know,

00:15:34.560 | why does it ever feel like anything to experience anything, right? And, you know, so I'm trying to

00:15:40.880 | distinguish from that problem, right? And say, you know, no, okay, I would merely settle for an

00:15:46.880 | account that could say, you know, is a fetus conscious? You know, if so, at which trimester?

00:15:52.560 | You know, is a dog conscious? You know, what about a frog, right?

00:15:58.160 | Or even as a precondition, you take that both these things are conscious,

00:16:02.080 | tell me which is more conscious.

00:16:03.680 | Yeah, for example, yes. Yeah, yeah. I mean, if consciousness is some multi-dimensional vector,

00:16:09.360 | well, just tell me in which respects these things are conscious and in which respect they aren't,

00:16:14.320 | right? And, you know, and have some principled way to do it where you're not, you know,

00:16:18.960 | carving out exceptions for things that you like or don't like, but could somehow take a description

00:16:24.800 | of an arbitrary physical system. And then just based on the physical properties of that system,

00:16:32.080 | or the informational properties, or how it's connected, or something like that,

00:16:36.800 | just in principle calculate, you know, its degree of consciousness, right? I mean, this would be the

00:16:43.040 | kind of thing that we would need, you know, if we wanted to address questions like, you know,

00:16:48.160 | what does it take for a machine to be conscious, right? Or when should we regard AIs as being

00:16:55.120 | conscious? So now this IIT, this integrated information theory, which has been put forward by

00:17:04.320 | Giulio Tononi and a bunch of his collaborators over the last decade or two, this is noteworthy,

00:17:14.880 | I guess, as a direct attempt to answer that question, to, you know, answer the, to address

00:17:21.040 | the pretty hard problem, right? And they give a criterion that's just based on how a system is

00:17:28.720 | connected. So you, so it's up to you to sort of abstract a system like a brain or a microchip

00:17:35.360 | as a collection of components that are connected to each other by some pattern of connections,

00:17:41.120 | you know, and to specify how the components can influence each other, you know, like where the

00:17:46.880 | inputs go, you know, where they affect the outputs. But then once you've specified that,

00:17:51.920 | then they give this quantity that they call phi, you know, the Greek letter phi.

00:17:56.880 | And the definition of phi has actually changed over time. It changes from one paper to another,

00:18:02.880 | but in all of the variations, it involves something about what we in computer science

00:18:08.640 | would call graph expansion. So basically what this means is that they want, in order to get

00:18:14.720 | a large value of phi, it should not be possible to take your system and partition it into two

00:18:22.160 | components that are only weakly connected to each other. Okay. So whenever we take our system and

00:18:28.800 | sort of try to split it up into two, then there should be lots and lots of connections going

00:18:33.600 | between the two components. Okay. Well, I understand what that means on a graph. Do they

00:18:38.000 | formalize what, how to construct such a graph or data structure, whatever, or is this, one of the

00:18:45.840 | criticism I've heard you kind of say is that a lot of the very interesting specifics are usually

00:18:51.840 | communicated through like natural language, like through words. So it's like the details aren't

00:18:58.320 | always clear. Well, it's true. I mean, they have nothing even resembling a derivation of this phi.

00:19:05.520 | Okay. So what they do is they state a whole bunch of postulates, you know, axioms that they think

00:19:11.600 | that consciousness should satisfy. And then there's some verbal discussion. And then at some

00:19:16.800 | point phi appears. Right. And this was the first thing that really made the hair stand on my neck,

00:19:23.520 | to be honest, because they are acting as if there's a derivation. They're acting as if, you

00:19:28.320 | know, you're supposed to think that this is a derivation and there's nothing even remotely

00:19:33.040 | resembling a derivation. They just pull the phi out of a hat completely. So is one of the key

00:19:38.400 | criticisms to you is that details are missing or is there something more fundamental? That's not

00:19:42.160 | even the key criticism. That's just a side point. Okay. The core of it is that I think that they

00:19:49.120 | want to say that a system is more conscious, the larger its value of phi. And I think that that is

00:19:55.120 | obvious nonsense. Okay. As soon as you think about it for like a minute, as soon as you think about

00:20:00.560 | it in terms of, could I construct a system that had an enormous value of phi, like, you know,

00:20:06.880 | even larger than the brain has, but that is just implementing an error correcting code, you know,

00:20:12.880 | doing nothing that we would associate with, you know, intelligence or consciousness or any of it.

00:20:18.880 | The answer is yes, it is easy to do that. Right. And so I wrote blog posts just making this point

00:20:24.800 | that, yeah, it's easy to do that. Now, you know, Tannone's response to that was actually kind of

00:20:29.920 | incredible. Right. I mean, I admired it in a way because instead of disputing any of it,

00:20:35.920 | he just bit the bullet in the sense, you know, he was one of the most audacious bullet bitings

00:20:43.120 | I've ever seen in my career. Okay. He said, okay, then fine. You know, this system that just applies

00:20:50.240 | this error correcting code, it's conscious, you know, and if it has a much larger value of phi,

00:20:55.360 | then you or me, it's much more conscious than you or me. You know, we just have to accept what the

00:21:01.040 | theory says because, you know, science is not about confirming our intuitions, it's about

00:21:06.080 | challenging them. And, you know, this is what my theory predicts, that this thing is conscious and,

00:21:11.680 | you know, or super duper conscious and how are you going to prove me wrong?

00:21:15.760 | See, so the way I would argue against your blog post is I would say, yes, sure, you're right in

00:21:22.240 | general, but for naturally arising systems developed through the process of evolution on Earth,

00:21:29.680 | this rule of the larger phi being associated with more consciousness is correct.

00:21:34.400 | Yeah, so that's not what he said at all, right? Because he wants this to be completely general.

00:21:39.680 | Even so we can apply to even computers.

00:21:41.760 | Yeah, I mean, the whole interest of the theory is the, you know, the hope that it could be

00:21:46.400 | completely general, apply to aliens, to computers, to animals, coma patients, to any of it, right?

00:21:54.560 | Yeah.

00:21:55.120 | And so he just said, well, you know, Scott is relying on his intuition, but, you know,

00:22:02.720 | I'm relying on this theory. And, you know, to me it was almost like, you know, are we being serious

00:22:08.400 | here? Like, you know, like, okay, yes, in science we try to learn highly non-intuitive things,

00:22:16.640 | but what we do is we first test the theory on cases where we already know the answer, right?

00:22:22.880 | Yeah.

00:22:23.120 | Like if someone had a new theory of temperature, right, then, you know, maybe we could check that

00:22:28.320 | it says that boiling water is hotter than ice. And then if it says that the sun is hotter than

00:22:34.240 | anything, you know, you've ever experienced, then maybe we trust that extrapolation, right?

00:22:40.480 | But like this theory, like if, you know, it's now saying that, you know, a gigantic grid,

00:22:48.320 | like regular grid of exclusive or gates can be way more conscious than a, you know, a person

00:22:54.800 | or than any animal can be, you know, even if it, you know, is, you know, is so uniform that it

00:23:02.160 | might as well just be a blank wall, right? And so now the point is, if this theory is sort of

00:23:07.920 | getting wrong, the question is a blank wall, you know, more conscious than a person, then I would

00:23:13.280 | say, what is there for it to get right?

00:23:15.920 | So your sense is a blank wall is not more conscious than a human being.

00:23:22.160 | Yeah, I mean, you could say that I am taking that as one of my axioms.

00:23:25.920 | I'm saying that if a theory of consciousness is getting that wrong, then whatever it is talking

00:23:35.040 | about at that point, I'm not going to call it consciousness. I'm going to use a different word.

00:23:40.640 | You have to use a different word. I mean, it's possible just like with intelligence

00:23:45.120 | that us humans conveniently define these very difficult to understand concepts in a very human

00:23:49.920 | centric way. Just like a Turing test really seems to define intelligence as a thing that's human-like.

00:23:56.320 | Right, but I would say that with any concept, you know, there's, you know, like we first need to

00:24:05.280 | define it, right? And a definition is only a good definition if it matches what we thought we were

00:24:10.800 | talking about, you know, prior to having a definition, right? And I would say that, you know,

00:24:15.600 | phi as a definition of consciousness fails that test. That is my argument.

00:24:22.400 | So, okay, so let's take a further step. So you mentioned that the universe might be

00:24:27.360 | the Turing machine, so like it might be computation.

00:24:30.800 | Or simulatable by one anyway.

00:24:32.560 | Simulatable by one. So what's your sense about consciousness? Do you think consciousness is

00:24:39.040 | computation? That we don't need to go to any place outside of the computable universe to,

00:24:46.240 | you know, to understand consciousness, to build consciousness, to measure consciousness,

00:24:52.720 | all those kinds of things?

00:24:54.080 | I don't know. These are what, you know, have been called the vertiginous questions, right?

00:24:59.840 | There's the questions like, you know, you get a feeling of vertigo when thinking about them,

00:25:05.120 | right? I mean, I certainly feel like I am conscious in a way that is not reducible to

00:25:11.680 | computation, but why should you believe me, right? I mean, and if you said the same to me,

00:25:17.680 | then why should I believe you?

00:25:19.360 | But as computer scientists, I feel like a computer could achieve human level intelligence.

00:25:26.320 | And that's actually a feeling and a hope. That's not a scientific belief. It's just we've built up

00:25:34.160 | enough intuition, the same kind of intuition you use in your blog. You know, that's what scientists

00:25:38.640 | do. I mean, some of it is a scientific method, but some of it is just damn good intuition.

00:25:42.880 | I don't have a good intuition about consciousness.

00:25:45.840 | Yeah. I'm not sure that anyone does or has in the, you know, 2,500 years that these things

00:25:51.600 | have been discussed, Lex.

00:25:52.640 | But do you think we will? Like one of the, I got a chance to attend, can't wait to hear your

00:25:58.800 | opinion on this, but attend the Neuralink event. And one of the dreams there is to, you know,

00:26:05.280 | basically push neuroscience forward. And the hope with neuroscience is that we can inspect

00:26:12.400 | the machinery from which all this fun stuff emerges and see, are we going to notice something

00:26:17.200 | special, some special sauce from which something like consciousness or cognition emerges?

00:26:22.720 | Yeah. Well, it's clear that we've learned an enormous amount about neuroscience.

00:26:27.040 | We've learned an enormous amount about computation, you know, about machine learning,

00:26:31.600 | about AI, how to get it to work. We've learned an enormous amount about the underpinnings of

00:26:38.560 | the physical world, you know. And, you know, from one point of view, that's like an enormous distance

00:26:45.120 | that we've traveled along the road to understanding consciousness. From another point of view,

00:26:49.920 | you know, the distance still to be traveled on the road, you know, maybe seems no shorter than

00:26:54.400 | it was at the beginning.

00:26:55.360 | Yeah.

00:26:55.760 | Right? So it's very hard to say. I mean, you know, these are questions like, in sort of trying to

00:27:02.480 | have a theory of consciousness, there's sort of a problem where it feels like it's not just that

00:27:07.520 | we don't know how to make progress, it's that it's hard to specify what could even count as progress.

00:27:13.360 | Right? Because no matter what scientific theory someone proposed, someone else could come along

00:27:18.160 | and say, "Well, you've just talked about the mechanism. You haven't said anything about

00:27:22.640 | what breathes fire into the mechanism, what really makes there something that it's like to be it."

00:27:27.920 | Right? And that seems like an objection that you could always raise, no matter, you know,

00:27:32.400 | how much someone elucidated the details of how the brain works.

00:27:35.840 | Okay, let's go to Turing test and Lobna Prize. I have this intuition, call me crazy,

00:27:40.000 | but we, that a machine to pass the Turing test in its full, whatever the spirit of it is,

00:27:48.000 | we can talk about how to formulate the perfect Turing test, that that machine has to be conscious.

00:27:54.880 | Or we at least have to, I have a very low bar of what consciousness is. I tend to think that

00:28:03.280 | the emulation of consciousness is as good as consciousness. So like consciousness is just a

00:28:08.640 | dance, a social shortcut, like a nice, useful tool. But I tend to connect intelligence and

00:28:16.240 | consciousness together. So by that, do you maybe just to ask, what role does consciousness play,

00:28:26.080 | do you think, in passing the Turing test?

00:28:27.520 | Well, look, I mean, it's almost tautologically true that if we had a machine that passed the

00:28:32.080 | Turing test, then it would be emulating consciousness, right? So if your position is

00:28:36.960 | that emulation of consciousness is consciousness, then by definition, any machine that passed the

00:28:43.920 | Turing test would be conscious. But I mean, you could say that that is just a way to rephrase the

00:28:51.200 | original question, is an emulation of consciousness necessarily conscious, right? And you can,

00:28:57.520 | I hear, I'm not saying anything new that hasn't been debated ad nauseum in the literature, okay?

00:29:03.840 | But you could imagine some very hard cases, like imagine a machine that passed the Turing test,

00:29:10.480 | but it did so just by an enormous cosmological-sized lookup table that just cached

00:29:17.200 | every possible conversation that could be had.

00:29:19.760 | The old Chinese room argument.

00:29:21.520 | Well, yeah, but this is, I mean, the Chinese room actually would be doing some computation,

00:29:27.360 | at least in Searle's version, right? Here, I'm just talking about a table lookup, okay? Now,

00:29:32.400 | it's true that for conversations of a reasonable length, this lookup table would be so enormous

00:29:38.080 | it wouldn't even fit in the observable universe, okay? But supposing that you could build a big

00:29:43.120 | enough lookup table and then just pass the Turing test just by looking up what the person said,

00:29:50.080 | right? Are you going to regard that as conscious?

00:29:52.960 | Okay, let me try to make this formal, and then you can shut it down. I think that the emulation

00:30:00.720 | of something is that something, if there exists in that system a black box that's full of mystery.

00:30:07.840 | So like...

00:30:08.240 | Full of mystery to whom?

00:30:11.440 | To human inspectors.

00:30:13.920 | So does that mean that consciousness is relative to the observer? Like, could something be conscious

00:30:18.400 | for us but not conscious for an alien that understood better what was happening inside

00:30:23.360 | the black box?

00:30:24.160 | Yes, yes. So that if inside the black box is just a lookup table, the alien that saw that would say

00:30:29.840 | this is not conscious. To us, another way to phrase the black box is layers of abstraction,

00:30:36.000 | which make it very difficult to see to the actually underlying functionality of the system.

00:30:40.560 | And then we observe just the abstraction, and so it looks like magic to us. But once we

00:30:45.680 | understand the inner machinery, it stops being magic. And so like, that's a prerequisite,

00:30:52.000 | is that you can't know how it works, some part of it. Because then there has to be, in our human mind,

00:30:58.160 | entry point for the magic. So that's a formal definition of the system.

00:31:05.520 | Yeah, well look, I mean, I explored a view in this essay I wrote called "The Ghost in the

00:31:10.640 | Quantum Turing Machine" seven years ago that is related to that, except that I did not want to

00:31:17.200 | have consciousness be relative to the observer, right? Because I think that if consciousness

00:31:22.080 | means anything, it is something that is experienced by the entity that is conscious, right? You know,

00:31:27.680 | like, I don't need you to tell me that I'm conscious, right? Nor do you need me to tell

00:31:34.720 | you that you are, right? But basically, what I explored there is, are there aspects of a system

00:31:44.160 | like a brain that just could not be predicted, even with arbitrarily advanced future technologies?

00:31:52.000 | It's because of chaos combined with quantum mechanical uncertainty, and things like that.

00:31:57.280 | I mean, that actually could be a property of the brain, if true, that would distinguish it in a

00:32:04.880 | principled way, at least from any currently existing computer. Not from any possible computer,

00:32:10.400 | but from, yeah, yeah. - Let's do a thought experiment.

00:32:12.640 | - Yeah. - So if I gave you

00:32:14.320 | information that you're in, the entire history of your life, basically explain away free will

00:32:22.640 | with a lookup table, say that this was all predetermined, that everything you experienced

00:32:27.200 | has already been predetermined, wouldn't that take away your consciousness? Wouldn't you,

00:32:31.520 | yourself, wouldn't the experience of the world change for you in a way that you can't take back?

00:32:37.920 | - Let me put it this way. If you could do like in a Greek tragedy, where you would just write down

00:32:43.520 | a prediction for what I'm going to do, and then maybe you put the prediction in a sealed box,

00:32:49.360 | and maybe you open it later, and you show that you knew everything I was going to do, or of course,

00:32:56.400 | the even creepier version would be you tell me the prediction, and then I try to falsify it,

00:33:01.680 | my very effort to falsify it makes it come true. Let's even forget that version, as convenient as

00:33:09.680 | it is for fiction writers. Let's just do the version where you put the prediction into a

00:33:14.320 | sealed envelope. But if you could reliably predict everything that I was going to do,

00:33:20.800 | I'm not sure that that would destroy my sense of being conscious, but I think it really would

00:33:25.600 | destroy my sense of having free will. And much, much more than any philosophical conversation

00:33:32.800 | could possibly do that. And so I think it becomes extremely interesting to ask, could such predictions

00:33:41.120 | be done, even in principle? Is it consistent with the laws of physics to make such predictions,

00:33:47.120 | to get enough data about someone that you could actually generate such predictions without having

00:33:52.160 | to kill them in the process, to slice their brain up into little slivers or something?

00:33:57.280 | >>ANDREW: I mean, theoretically possible, right?

00:33:58.960 | >>BETSY: Well, I don't know. I mean, it might be possible, but only at the cost of destroying the

00:34:03.280 | person, right? I mean, it depends on how low you have to go in sort of the substrate. Like,

00:34:10.320 | if there was a nice digital abstraction layer, if you could think of each neuron as a kind of

00:34:16.080 | transistor computing a digital function, then you could imagine some nanorobots that would go in

00:34:22.160 | and would just scan the state of each transistor, of each neuron, and then make a good enough copy.

00:34:29.440 | But if it was actually important to get down to the molecular or the atomic level,

00:34:35.520 | then eventually you would be up against quantum effects. You would be up against the unclone

00:34:40.320 | ability of quantum states. So I think it's a question of how good does the replica have to be

00:34:48.160 | before you're going to count it as actually a copy of you or as being able to predict your actions.

00:34:53.600 | >>ANDREW: And that's a totally open question.

00:34:55.520 | >>BETSY: Yeah, yeah, yeah. And especially once we say that, well, look, maybe there's no way

00:35:02.400 | to make a deterministic prediction because we know that there's noise buffeting the brain around,

00:35:09.120 | presumably even quantum mechanical uncertainty affecting the sodium ion channels, for example,

00:35:15.360 | whether they open or they close. There's no reason why over a certain timescale that shouldn't be

00:35:23.680 | amplified, just like we imagine happens with the weather or with any other chaotic system.

00:35:31.920 | So if that stuff is important, then we would say, well, you're never going to be able to make an

00:35:44.560 | accurate enough copy. But now the hard part is, well, what if someone can make a copy that no one

00:35:50.080 | else can tell apart from you? It says the same kinds of things that you would have said, maybe

00:35:56.240 | not exactly the same things because we agree that there's noise, but it says the same kinds of

00:36:01.280 | things. And maybe you alone would say, no, I know that that's not me. I haven't felt my consciousness

00:36:09.520 | leap over to that other thing. I still feel it localized in this version, right? Then why should

00:36:14.960 | anyone else believe you? >>ENOCH: What are your thoughts? I'd be curious. You're a really good

00:36:19.120 | person to ask, which is Penrose's, Roger Penrose's work on consciousness, saying that there is some

00:36:26.160 | - with axons and so on - there might be some biological places where quantum mechanics can

00:36:32.240 | come into play and through that create consciousness somehow. >>ANDREW: Yeah. Okay.

00:36:36.480 | >>ENOCH: Are you familiar with his work at all? >>ANDREW: Of course. I read Penrose's books as a

00:36:41.280 | teenager. They had a huge impact on me. Five or six years ago, I had the privilege to actually

00:36:47.200 | talk these things over with Penrose at some length at a conference in Minnesota. And he is

00:36:54.800 | an amazing personality. I admire the fact that he was even raising such audacious questions at all.

00:37:01.600 | But to answer your question, I think the first thing we need to get clear on is that he is not

00:37:08.640 | merely saying that quantum mechanics is relevant to consciousness. That would be tame compared to

00:37:16.320 | what he is saying. He is saying that even quantum mechanics is not good enough. Because if supposing

00:37:23.680 | for example that the brain were a quantum computer, well, that's still a computer. In fact,

00:37:28.720 | a quantum computer can be simulated by an ordinary computer. It might merely need exponentially more

00:37:34.880 | time in order to do so. So that's simply not good enough for him. So what he wants is for the brain

00:37:41.680 | to be a quantum gravitational computer. Or he wants the brain to be exploiting as-yet-unknown

00:37:50.960 | laws of quantum gravity, which would be uncomputable according to him.

00:37:56.000 | That's the key point.

00:37:56.880 | Okay, yes, yes. That would be literally uncomputable. And I've asked him to clarify

00:38:02.400 | this. But uncomputable, even if you had an oracle for the halting problem, or as high up as you want

00:38:11.280 | to go in the usual hierarchy of uncomputability, he wants to go beyond all of that. So just to be

00:38:20.000 | clear, if we're keeping count of how many speculations, there's probably at least five

00:38:26.160 | or six of them, right? There's first of all that there is some quantum gravity theory that would

00:38:30.960 | involve this kind of uncomputability, right? Most people who study quantum gravity would not agree

00:38:36.560 | with that. They would say that what we've learned, what little we know about quantum gravity from

00:38:42.160 | this ADS-CFT correspondence, for example, has been very much consistent with the broad idea

00:38:49.200 | of nature being computable, right? But supposing that he's right about that, then what most

00:38:58.720 | physicists would say is that whatever new phenomena there are in quantum gravity, they might be

00:39:05.520 | relevant at the singularities of black holes. They might be relevant at the Big Bang. They are

00:39:13.280 | plainly not relevant to something like the brain that is operating at ordinary temperatures

00:39:19.760 | with ordinary chemistry and the physics underlying the brain. They would say that we have the

00:39:29.440 | fundamental physics of the brain, they would say that we've pretty much completely known for

00:39:34.720 | generations now, right? Because quantum field theory lets us sort of parametrize our ignorance,

00:39:42.160 | right? I mean, Sean Carroll has made this case in great detail, right? That sort of whatever

00:39:47.920 | new effects are coming from quantum gravity, they are sort of screened off by quantum field theory,

00:39:54.080 | right? And this brings us to the whole idea of effective theories, right? But we have,

00:40:00.400 | like in the standard model of elementary particles, right? We have a quantum field theory

00:40:07.280 | that seems totally adequate for all of the terrestrial phenomena, right? The only things

00:40:13.360 | that it doesn't explain are, well, first of all, the details of gravity, if you were to probe it

00:40:19.680 | at extremes of curvature or at incredibly small distances. It doesn't explain dark matter. It

00:40:28.880 | doesn't explain black hole singularities, right? But these are all very exotic things, very far

00:40:34.800 | removed from our life on Earth, right? So for Penrose to be right, he needs these phenomena to

00:40:41.600 | somehow affect the brain. He needs the brain to contain antennae that are sensitive to this--

00:40:47.920 | >> Black holes.

00:40:48.400 | >> --to this as-yet-unknown physics, right? And then he needs a modification of quantum mechanics,

00:40:55.040 | okay? So he needs quantum mechanics to actually be wrong, okay? He needs what he wants is what

00:41:02.160 | he calls an objective reduction mechanism or an objective collapse. So this is the idea that once

00:41:08.560 | quantum states get large enough, then they somehow spontaneously collapse, right? And this is an idea

00:41:18.960 | that lots of people have explored. There's something called the GRW proposal that tries to

00:41:26.080 | say something along those lines. And these are theories that actually make testable predictions,

00:41:31.920 | right? Which is a nice feature that they have. But the very fact that they're testable may mean

00:41:36.640 | that in the coming decades, we may well be able to test these theories and show that they're wrong,

00:41:43.840 | right? We may be able to test some of Penrose's ideas. If not, not his ideas about consciousness,

00:41:50.480 | but at least his ideas about an objective collapse of quantum states, right? And people have actually,

00:41:56.560 | like Dick Balmister, have actually been working to try to do these experiments.

00:42:00.880 | They haven't been able to do it yet to test Penrose's proposal, okay? But Penrose would

00:42:05.840 | need more than just an objective collapse of quantum states, which would already be the

00:42:10.800 | biggest development in physics for a century since quantum mechanics itself, okay? He would need

00:42:16.160 | for consciousness to somehow be able to influence the direction of the collapse so that it wouldn't

00:42:23.520 | be completely random, but that your dispositions would somehow influence the quantum state to

00:42:29.680 | collapse more likely this way or that way, okay? Finally, Penrose says that all of this has to be

00:42:37.200 | true because of an argument that he makes based on Gödel's incompleteness theorem, okay?

00:42:42.560 | >>Right.

00:42:43.120 | >>Now, like, I would say, the overwhelming majority of computer scientists and mathematicians

00:42:49.120 | who have thought about this, I don't think that Gödel's incompleteness theorem can do what he

00:42:54.000 | needs it to do here, right? I don't think that that argument is sound, okay? But that is sort of

00:43:00.720 | the tower that you have to ascend to if you're going to go where Penrose goes.

00:43:04.560 | >>And the intuition he uses with the incompleteness theorem is that basically

00:43:09.440 | that there's important stuff that's not computable? Is that where he takes it?

00:43:13.120 | >>No, it's not just that, because, I mean, everyone agrees that there are problems that

00:43:17.120 | are uncomputable, right? That's a mathematical theorem, right? But what Penrose wants to say

00:43:22.960 | is that, for example, there are statements, given any formal system for doing math, right? There

00:43:34.400 | will be true statements of arithmetic that that formal system, if it's adequate for math at all,

00:43:41.200 | if it's consistent and so on, will not be able to prove. A famous example being the statement

00:43:46.960 | that that system itself is consistent, right? No good formal system can actually prove its own

00:43:53.600 | consistency. That can only be done from a stronger formal system, which then can't prove its own

00:43:59.600 | consistency and so on forever, okay? That's Gödel's theorem. But now, why is that relevant to

00:44:06.080 | consciousness, right? Well, I mean, the idea that it might have something to do with consciousness

00:44:13.360 | is an old one. Gödel himself apparently thought that it did. Lucas thought so, I think, in the 60s.

00:44:23.360 | And Penrose is really just sort of updating what they and others had said. I mean, the idea that

00:44:30.640 | Gödel's theorem could have something to do with consciousness was in 1950, when Alan Turing wrote

00:44:37.440 | his article about the Turing test. He already was writing about that as an old and well-known idea

00:44:44.720 | and as a wrong one that he wanted to dispense with, right? Okay, but the basic problem with

00:44:51.760 | this idea is Penrose wants to say that, and all of his predecessors here want to say, that even

00:44:59.600 | though this given formal system cannot prove its own consistency, we as humans, sort of looking at

00:45:07.520 | it from the outside, can just somehow see its consistency, right? And the rejoinder to that

00:45:15.760 | from the very beginning has been, "Well, can we really?" I mean, maybe he, Penrose, can,

00:45:23.680 | but can the rest of us? And notice that it is perfectly plausible to imagine a computer

00:45:35.200 | that could say, would not be limited to working within a single formal system, right? They could

00:45:41.120 | say, "I am now going to adopt the hypothesis that my formal system is consistent, and I'm now going

00:45:47.840 | to see what can be done from that stronger vantage point," and so on. And I'm going to add new

00:45:53.280 | axioms to my system. Totally plausible. There's absolutely, Godel's theorem has nothing to say

00:45:59.520 | against an AI that could repeatedly add new axioms. All it says is that there is no

00:46:05.920 | absolute guarantee that when the AI adds new axioms that it will always be right.

00:46:11.760 | Okay? And that's, of course, the point that Penrose pounces on, but the reply is obvious.

00:46:17.280 | And it's one that Alan Turing made 70 years ago, namely, we don't have an absolute guarantee that

00:46:22.880 | we're right when we add a new axiom. We never have, and plausibly we never will.

00:46:27.840 | So on Alan Turing, you took part in the Lubna Prize?

00:46:31.600 | Not really, no.

00:46:33.680 | Or was it?

00:46:34.240 | I didn't. I mean, there was this kind of ridiculous claim that was made almost a decade ago about a

00:46:43.280 | chatbot called Eugene Gooseman.

00:46:45.360 | Yeah, I apologize. I guess you didn't participate as a judge in the Lubna Prize.

00:46:48.640 | I didn't, no.

00:46:49.120 | But you participated as a judge in that, I guess it was an exhibition event or something like that.

00:46:54.240 | Or was Eugene the...

00:46:55.440 | Eugene Gooseman, that was just me writing a blog post,

00:46:59.360 | because some journalist called me to ask about it.

00:47:01.680 | Did you ever chat with him?

00:47:03.200 | I did chat with Eugene Gooseman. I mean, it was available on the web, the chat.

00:47:06.640 | Oh, interesting. I didn't know that.

00:47:07.680 | So yeah. So all that happened was that a bunch of journalists started writing breathless articles

00:47:14.400 | about first chatbot that passes the Turing test. And it was this thing called Eugene Gooseman

00:47:21.440 | that was supposed to simulate a 13-year-old boy. And apparently someone had done some test where

00:47:29.120 | people were less than perfect, let's say, distinguishing it from a human.

00:47:35.440 | And they said, well, if you look at Turing's paper and you look at the percentages that he

00:47:41.600 | talked about, then it seems like we're past that threshold. And I had a different way to look at it

00:47:50.400 | instead of the legalistic way, like, let's just try the actual thing out and let's see what it can do

00:47:56.480 | with questions like, is Mount Everest bigger than a shoebox? Or just the most obvious questions.

00:48:04.560 | And the answer is, well, it just parries you because it doesn't know what you're talking about.

00:48:11.200 | So just to clarify exactly in which way they're obvious. They're obvious in the sense that you

00:48:17.840 | convert the sentences into the meaning of the objects they represent and then do some basic

00:48:22.960 | obvious, we mean, common sense reasoning with the objects that the sentences represent.

00:48:29.200 | Right, right. It was not able to answer or even intelligently respond to basic common

00:48:34.720 | sense questions. But let me say something stronger than that. There was a famous chatbot in the '60s

00:48:39.920 | called Eliza that managed to actually fool a lot of people. People would pour their hearts out into

00:48:48.400 | this Eliza because it simulated a therapist. And most of what it would do is it would just throw

00:48:54.320 | back at you whatever you said. And this turned out to be incredibly effective. Maybe therapists

00:49:02.400 | know this. This is one of their tricks. But it really had some people convinced.

00:49:10.240 | But this thing was just like, I think it was literally just a few hundred lines of Lisp code.

00:49:16.880 | Not only was it not intelligent, it wasn't especially sophisticated. It was a simple

00:49:24.080 | little hobbyist program. And Eugene Guzman, from what I could see, was not a significant advance

00:49:29.920 | compared to Eliza. And that was really the point I was making. And in some sense, you didn't need a

00:49:40.160 | computer science professor to say this. Anyone who was looking at it and who just had an ounce of

00:49:48.880 | sense could have said the same thing. But because these journalists were calling me, the first thing

00:49:56.800 | I said was, "Well, no, I'm a quantum computing person. I'm not an AI person. You shouldn't ask

00:50:03.760 | me." Then they said, "Look, you can go here and you can try it out." I said, "All right, all right,

00:50:07.680 | so I'll try it out." But now this whole discussion, it got a whole lot more interesting in just the

00:50:14.720 | last few months. Yeah, I'd love to hear your thoughts about GPT-3, the advancements in length.

00:50:19.280 | In the last few months, the world has now seen a chat engine or a text engine, I should say,

00:50:28.000 | called GPT-3. I think it does not pass a Turing test. There are no real claims that it passes

00:50:37.680 | the Turing test. This comes out of the group at OpenAI and they've been relatively careful in

00:50:45.200 | what they've claimed about the system. But I think this, as clearly as Eugene Guzman was not

00:50:53.760 | an advance over ELISA, it is equally clear that this is a major advance over ELISA or really over

00:51:01.280 | anything that the world has seen before. This is a text engine that can come up with on-topic,

00:51:10.320 | reasonable-sounding completions to just about anything that you ask. You can ask it to write

00:51:16.000 | a poem about topic X in the style of poet Y, and it will have a go at that. And it will do

00:51:23.520 | not a great job, not an amazing job, but a passable job. Definitely as good as,

00:51:33.760 | in many cases, I would say better than I would have done. You can ask it to write an essay,

00:51:40.240 | like a student essay about pretty much any topic, and it will get something that I am pretty sure

00:51:45.680 | would get at least a B- in most high school or even college classes. In some sense, the way that

00:51:55.360 | it achieves this, Scott Alexander of the much mourned blog Slate Star Codex had a wonderful

00:52:04.000 | way of putting it. He said that they basically just ground up the entire internet into a slurry.

00:52:09.680 | To tell you the truth, I had wondered for a while why nobody had tried that. Why not write a chat

00:52:18.400 | bot by just doing deep learning over a corpus consisting of the entire web? Now they finally

00:52:28.160 | have done that. The results are very impressive. It's not clear that people can argue about whether

00:52:36.640 | this is truly a step toward general AI or not, but this is an amazing capability that we didn't

00:52:45.440 | have a few years ago. A few years ago, if you had told me that we would have it now, that would have

00:52:51.840 | surprised me. I think that anyone who denies that is just not engaging with what's there.

00:52:57.280 | Their model takes a large part of the internet and compresses it in a small number of parameters

00:53:05.200 | relative to the size of the internet, and is able to, without fine tuning,

00:53:12.320 | do a basic kind of a querying mechanism, just like you describe when you specify a kind of poet,

00:53:18.240 | and then you want to write a poem. It somehow is able to do basically a lookup on the internet

00:53:23.120 | of relevant things. How else do you explain it?

00:53:27.440 | Well, okay. The training involved massive amounts of data from the internet, and actually took

00:53:34.080 | lots and lots of computer power, lots of electricity. There are some very prosaic

00:53:40.080 | reasons why this wasn't done earlier, right? But it costs some tens of millions of dollars,

00:53:46.000 | I think.

00:53:46.400 | Less, but approximately a few million dollars.

00:53:49.520 | Oh, okay, okay. Oh, really? Okay.

00:53:51.440 | It's more like four or five.

00:53:53.680 | Oh, all right, all right. Thank you. I mean, as they scale it up, it will-

00:53:57.440 | It'll cost. But then the hope is cost comes down, and all that kind of stuff.

00:54:01.600 | But basically, it is a neural net, or what's now called a deep net, but they're

00:54:09.760 | basically the same thing, right? So it's a form of algorithm that people have known about for

00:54:16.400 | decades, right? But it is constantly trying to solve the problem, predict the next word,

00:54:23.680 | right? So it's just trying to predict what comes next. It's not trying to decide what it should

00:54:32.160 | say, what ought to be true. It's trying to predict what someone who had said all of the words up to

00:54:38.320 | the preceding one would say next.

00:54:40.720 | Although to push back on that, that's how it's trained, but-

00:54:43.520 | That's right. No, of course.

00:54:44.240 | But it's arguable that our very cognition could be a mechanism as that simple.

00:54:50.560 | Oh, of course. Of course. I never said that it wasn't.

00:54:52.880 | Right. But, yeah.

00:54:55.200 | In some sense, if there is a deep philosophical question that's raised by GPT-3, then that is it,

00:55:02.800 | right? Are we doing anything other than this predictive processing,

00:55:06.960 | just constantly trying to fill in a blank of what would come next after what we just said up to this

00:55:13.680 | point? Is that what I'm doing right now? Is it possible, so the intuition that a lot

00:55:19.280 | of people have, "Well, look, this thing is not going to be able to reason, the Mountain Everest

00:55:24.000 | question." Do you think it's possible that GPT-5, 6, and 7 would be able to, with this exact same

00:55:31.040 | process, begin to do something that looks like is indistinguishable to us humans from reasoning?

00:55:38.720 | I mean, the truth is that we don't really know what the limits are, right?

00:55:42.960 | Right. Exactly.

00:55:44.000 | Because what we've seen so far is that GPT-3 was basically the same thing as GPT-2, but just with

00:55:52.400 | a much larger network, more training time, bigger training corpus, and it was very noticeably better

00:56:02.960 | than its immediate predecessor. We don't know where you hit the ceiling here. I mean, that's

00:56:09.920 | the amazing part and maybe also the scary part. Now, my guess would be that at some point,

00:56:18.560 | there has to be diminishing returns. It can't be that simple, can it? Right? But I wish that I had

00:56:25.840 | more to base that guess on. Right. Yeah. I mean, some people say that there

00:56:29.760 | would be a limitation on the, "We're going to hit a limit on the amount of data that's

00:56:33.760 | on the internet." Yes. Yeah. So, sure. So, there's certainly that limit. I mean, there's also,

00:56:39.360 | if you are looking for questions that will stump GPT-3, you can come up with some without,

00:56:48.320 | even getting it to learn how to balance parentheses. Right? It doesn't do such a great

00:56:55.200 | job. Right? Its failures are ironic, right? Like basic arithmetic. Right? And you think,

00:57:05.280 | "Isn't that what computers are supposed to be best at? Isn't that where computers already had

00:57:09.760 | us beat a century ago?" Right? And yet, that's where GPT-3 struggles. Right? But it's amazing

00:57:16.640 | that it's almost like a young child in that way. Right? But somehow, because it is just trying to

00:57:25.600 | predict what comes next, it doesn't know when it should stop doing that and start doing something

00:57:32.400 | very different, like some more exact logical reasoning. Right? And so, one is naturally led

00:57:41.520 | to guess that our brain sort of has some element of predictive processing, but that it's coupled to

00:57:48.080 | other mechanisms. Right? That it's coupled to, first of all, visual reasoning, which GPT-3 also

00:57:53.920 | doesn't have any of. Right? Although there's some demonstration that there's a lot of promise there.

00:57:57.920 | Oh, yeah. It can complete images. That's right. Yeah. And using exact same kind of transformer

00:58:03.200 | mechanisms to watch videos on YouTube. And so, the same self-supervised mechanism to be able to

00:58:11.360 | it'd be fascinating to think what kind of completions you could do.

00:58:14.240 | Oh, yeah. No, absolutely. Although, if we ask it to a word problem that involved reasoning about

00:58:20.400 | the locations of things in space, I don't think it does such a great job on those. Right? To take an

00:58:25.440 | example. And so, the guess would be, well, humans have a lot of predictive processing, a lot of just

00:58:31.840 | filling in the blanks, but we also have these other mechanisms that we can couple to, or that we can

00:58:37.360 | call a subroutines when we need to. And that maybe to go further that one would want to integrate

00:58:44.480 | other forms of reasoning. Let me go on another topic that is amazing, which is complexity.

00:58:52.320 | And then start with the most absurdly romantic question of what's the most beautiful idea in

00:59:00.560 | the computer science or theoretical computer science to you? Like what just early on in your

00:59:05.520 | life or in general have captivated you and just grabbed you?

00:59:08.560 | I think I'm going to have to go with the idea of universality. If you're really asking for the most

00:59:14.800 | beautiful. I mean, so universality is the idea that you put together a few simple operations.

00:59:23.040 | In the case of Boolean logic, that might be the AND gate, the OR gate, the NOT gate. And then

00:59:30.560 | your first guess is, okay, this is a good start, but obviously, as I want to do more complicated

00:59:36.800 | things, I'm going to need more complicated building blocks to express that. And that was actually my

00:59:42.560 | guess when I first learned what programming was. I mean, when I was an adolescent and someone showed

00:59:48.480 | me Apple Basic and GW Basic, if anyone listening remembers that. But I thought, okay, well now,

01:00:00.000 | I felt like this is a revelation. It's like finding out where babies come from. It's like

01:00:06.480 | that level of, why didn't anyone tell me this before? But I thought, okay, this is just the

01:00:11.520 | beginning. Now I know how to write a basic program, but to really write an interesting program,

01:00:18.080 | like a video game, which had always been my dream as a kid to create my own Nintendo games.

01:00:25.120 | But obviously, I'm going to need to learn some way more complicated form of programming than that.

01:00:30.800 | But eventually I learned this incredible idea of universality. And that says that, no,

01:00:38.560 | you throw in a few rules and then you already have enough to express everything. So for example,

01:00:45.840 | the AND, the OR, and the NOT gate can all, or in fact, even just the AND and the NOT gate,

01:00:51.600 | or even just the NAND gate, for example, is already enough to express any Boolean function

01:00:58.160 | on any number of bits. You just have to string together enough of them.

01:01:01.600 | You can build a universe with NAND gates.

01:01:03.440 | You can build the universe out of NAND gates. Yeah. The simple instructions of Basic are already

01:01:10.000 | enough, at least in principle. If we ignore details like how much memory can be accessed

01:01:16.560 | and stuff like that, that is enough to express what could be expressed by any programming language

01:01:21.760 | whatsoever. And the way to prove that is very simple. We simply need to show that in Basic,

01:01:27.680 | or whatever, we could write an interpreter or a compiler for whatever other programming language

01:01:34.160 | we care about, like C or Java or whatever. And as soon as we had done that, then ipso facto,

01:01:40.720 | anything that's expressible in C or Java is also expressible in Basic.

01:01:45.200 | And so this idea of universality goes back at least to Alan Turing in the 1930s, when he

01:01:54.720 | wrote down this incredibly simple pared-down model of a computer, the Turing machine, which

01:02:04.080 | he pared down the instruction set to just read a symbol, write a symbol, move to the left,

01:02:11.760 | move to the right, halt, change your internal state. That's it. And he proved that this could

01:02:21.120 | simulate all kinds of other things. And so in fact, today we would say, well, we would call it

01:02:27.920 | a Turing universal model of computation that has just the same expressive power that Basic or

01:02:36.320 | Java or C++ or any of those other languages have, because anything in those other languages could

01:02:44.080 | be compiled down to Turing machine. Now, Turing also proved a different related thing,

01:02:50.000 | which is that there is a single Turing machine that can simulate any other Turing machine.

01:02:56.800 | If you just describe that other machine on its tape, right? And likewise, there is a single

01:03:03.360 | Turing machine that will run any C program, you know, if you just put it on its tape. That's

01:03:08.480 | a second meaning of universality. First of all, he couldn't visualize it,

01:03:13.200 | and that was in the 30s? Yeah, the 30s, that's right.

01:03:15.520 | That's before computers really... I mean, I don't know how... I wonder what that felt like,

01:03:24.480 | you know, learning that there's no Santa Claus or something. Because I don't know if that's

01:03:30.000 | empowering or paralyzing, because it doesn't give you any... It's like, you can't write a

01:03:36.720 | software engineering book and make that the first chapter and say, we're done.

01:03:40.400 | Well, I mean, right. I mean, in one sense, it was this enormous flattening of the universe, right?

01:03:47.520 | I had imagined that there was going to be some infinite hierarchy of more and more powerful

01:03:52.880 | programming languages. And then I kicked myself for having such a stupid idea, but apparently

01:03:58.880 | Gödel had had the same conjecture in the 30s. Oh, good. You're in good company.

01:04:03.200 | Yeah, and then Gödel read Turing's paper, and he kicked himself, and he said, "Yeah,

01:04:09.360 | I was completely wrong about that." Okay, but I had thought that maybe where I can contribute will

01:04:16.560 | be to invent a new, more powerful programming language that lets you express things that could

01:04:21.760 | never be expressed in basic. And how would you do that? Obviously, you couldn't do it itself in

01:04:27.760 | basic, right? But there is this incredible flattening that happens once you learn what

01:04:34.000 | is universality. But then it's also an opportunity, because it means once you know these rules,

01:04:42.560 | then the sky is the limit, right? Then you have the same weapons at your disposal that the world's

01:04:49.840 | greatest programmer has. It's now all just a question of how you wield them.

01:04:54.320 | Right, exactly. So every problem is solvable, but some problems are harder than others.

01:05:00.960 | Well, yeah, there's the question of how much time, of how hard is it to write a program?

01:05:06.960 | And then there's also the questions of what resources does the program need? How much time,

01:05:12.080 | how much memory? Those are much more complicated questions, of course, ones that we're still

01:05:16.080 | struggling with today. Exactly. So you've, I don't know if you created Complexity Zoo, or...

01:05:21.200 | I did create the Complexity Zoo. What is it? What's complexity?

01:05:24.960 | Oh, all right, all right, all right. Complexity theory is the study of the inherent resources

01:05:31.360 | needed to solve computational problems. So it's easiest to give an example. Let's say we want to

01:05:41.840 | add two numbers, right? If I want to add them, if the numbers are twice as long, then it will take

01:05:49.440 | me twice as long to add them, but only twice as long, right? It's no worse than that.

01:05:54.480 | For a computer.

01:05:55.440 | For a computer, or for a person, using pencil and paper for that matter.

01:05:59.120 | If you have a good algorithm.

01:06:00.400 | Yeah, that's right. I mean, even if you just use the elementary school algorithm of just carrying,

01:06:05.440 | you know, then it takes time that is linear in the length of the numbers, right? Now,

01:06:10.640 | multiplication, if you use the elementary school algorithm, is harder because you have to multiply

01:06:17.040 | each digit of the first number by each digit of the second one, and then deal with all the carries.

01:06:22.720 | So that's what we call a quadratic time algorithm, right? If the numbers become twice as long,

01:06:29.120 | now you need four times as much time, okay? So now, as it turns out, people discovered much

01:06:38.000 | faster ways to multiply numbers using computers. And today we know how to multiply two numbers that

01:06:45.200 | are n digits long using a number of steps that's nearly linear in n. These are questions you can

01:06:51.120 | ask, but now let's think about a different thing that people, you know, have encountered in

01:06:56.320 | elementary school, factoring a number. Okay, take a number and find its prime factors, right?

01:07:03.040 | And here, you know, if I give you a number with 10 digits, I ask you for its prime factors,

01:07:08.640 | well, maybe it's even, so you know that two is a factor. You know, maybe it ends in zero,

01:07:13.600 | so you know that 10 is a factor, right? But, you know, other than a few obvious things like that,

01:07:18.880 | you know, if the prime factors are all very large, then it's not clear how you even get started,

01:07:24.320 | right? You know, it seems like you have to do an exhaustive search among an enormous number

01:07:29.200 | of factors. Now, and as many people might know, for better or worse, the security, you know,

01:07:39.280 | of most of the encryption that we currently use to protect the internet is based on the belief,

01:07:45.200 | and this is not a theorem, it's a belief, that factoring is an inherently hard problem

01:07:52.080 | for our computers. We do know algorithms that are better than just trial division,

01:07:57.360 | just trying all the possible divisors, but they are still basically exponential.

01:08:03.520 | >> And exponential is hard.

01:08:05.280 | >> Yeah, exactly. So the fastest algorithms that anyone has discovered,

01:08:10.560 | at least publicly discovered, you know, I'm assuming that the NSA doesn't know something

01:08:14.800 | better, okay, but they take time that basically grows exponentially with the cube root of the

01:08:21.520 | size of the number that you're factoring, right? So that cube root, that's the part that takes all

01:08:26.720 | the cleverness, okay, but there's still an exponential, there's still an exponentiality

01:08:31.040 | there. What that means is that, like, when people use a thousand bit keys for their cryptography,

01:08:37.360 | that can probably be broken using the resources of the NSA or the world's other intelligence

01:08:42.800 | agencies. You know, people have done analyses that say, you know, with a few hundred million

01:08:47.600 | dollars of computer power, they could totally do this. And if you look at the documents that

01:08:52.480 | Snowden released, you know, it looks a lot like they are doing that or something like that. It

01:08:59.040 | would kind of be surprising if they weren't, okay? But, you know, if that's true, then in some ways,

01:09:05.200 | that's reassuring, because if that's the best that they can do, then that would say that they

01:09:09.360 | can't break 2000-bit numbers, right?

01:09:12.160 | >> Right, exactly.

01:09:12.960 | >> Right, then 2000-bit numbers would be beyond what even they could do.

01:09:16.880 | >> They haven't found an efficient algorithm. That's where all the worries and the concerns of

01:09:21.360 | quantum computing came in, that there could be some kind of shortcut around that.

01:09:24.160 | >> Right. So, complexity theory is a, you know, is a huge part of, let's say, the theoretical core

01:09:31.280 | of computer science. You know, it started in the '60s and '70s as, you know, sort of an, you know,

01:09:38.080 | autonomous field. So it was, you know, already, you know, I mean, you know, it was well-developed

01:09:43.840 | even by the time that I was born, okay? But I, in 2002, I made a website called The Complexity Zoo,

01:09:53.040 | to answer your question, where I just tried to catalog the different complexity classes,

01:09:59.520 | which are classes of problems that are solvable with different kinds of resources, okay?

01:10:04.800 | So these are kind of, you know, you could think of complexity classes as like being almost to

01:10:11.680 | theoretical computer science, like what the elements are to chemistry, right? They're sort of,

01:10:16.560 | you know, they're our most basic objects in a certain way.

01:10:20.720 | >> I feel like the elements have a characteristic to them where you can't just add an infinite number.

01:10:28.880 | >> Well, you could, but beyond a certain point, they become unstable, right? So it's like, you

01:10:35.040 | know, in theory, you can have atoms with, you know, and look, look, I mean, a neutron star,

01:10:40.400 | you know, is a nucleus with, you know, unculled billions of neutrons in it, of hadrons in it,

01:10:50.080 | okay? But, you know, for sort of normal atoms, right, probably you can't get much above 100,

01:10:57.600 | you know, atomic weight, 150 or so, or sorry, sorry, I mean, beyond 150 or so protons without

01:11:05.360 | very quickly fissioning. With complexity classes, well, yeah, you can have an infinity of complexity

01:11:11.680 | classes. But, you know, maybe there's only a finite number of them that are particularly

01:11:16.880 | interesting, right? Just like with anything else, you know, you care about some more than about

01:11:22.640 | others.

01:11:22.960 | >> So what kind of interesting classes are there? I mean, you could have just maybe say,

01:11:28.000 | what are the, if you take any kind of computer science class, what are the classes you learn?

01:11:32.400 | >> Good. Let me tell you sort of the biggest ones, the ones that you would learn first. So,

01:11:38.160 | you know, first of all, there is P, that's what it's called, okay? It stands for polynomial time.

01:11:44.400 | And this is just the class of all of the problems that you could solve with a conventional computer,

01:11:50.960 | like your iPhone or your laptop, you know, by a completely deterministic algorithm,

01:11:57.120 | right? Using a number of steps that grows only like the size of the input raised to some fixed

01:12:04.320 | power, okay? So if your algorithm is linear time, like, you know, for adding numbers, okay, that

01:12:11.920 | problem is in P. If you have an algorithm that's quadratic time, like the elementary school

01:12:17.920 | algorithm for multiplying two numbers, that's also in P. Even if it was the size of the input

01:12:23.200 | to the 10th power or to the 50th power, well, that wouldn't be very good in practice. But,

01:12:29.360 | you know, formally, we would still count that. That would still be in P, okay? But if your

01:12:34.000 | algorithm takes exponential time, meaning like if every time I add one more data point to your input,

01:12:43.280 | if the time needed by the algorithm doubles, if you need time like two to the power of the amount

01:12:50.480 | of input data, then that we call an exponential time algorithm, okay? And that is not polynomial,

01:12:58.000 | okay? So P is all of the problems that have some polynomial time algorithm, okay?

01:13:04.720 | So that includes most of what we do with our computers on a day-to-day basis, you know,

01:13:09.760 | all the, you know, sorting, basic arithmetic, you know, whatever is going on in your email reader

01:13:15.600 | or in Angry Birds, okay? It's all in P. Then the next super important class is called NP.

01:13:23.040 | That stands for non-deterministic polynomial, okay? Does not stand for not polynomial, which is a

01:13:30.240 | common confusion. But NP was basically all of the problems where if there is a solution,

01:13:38.000 | then it is easy to check the solution if someone shows it to you, okay? So actually a perfect

01:13:43.760 | example of a problem in NP is factoring, the one I told you about before. Like if I gave you a

01:13:51.360 | number with thousands of digits and I told you that, you know, I asked you, "Does this have at

01:13:58.800 | least three non-trivial divisors?" Right? That might be a super hard problem to solve, right?

01:14:05.920 | Might take you millions of years using any algorithm that's known, at least running on

01:14:10.640 | our existing computers, okay? But if I simply showed you the divisors, I said, "Here are three

01:14:17.040 | divisors of this number," then it would be very easy for you to ask your computer to just check

01:14:22.800 | each one and see if it works. Just divide it in, see if there's any remainder, right? And if they

01:14:28.240 | all go in, then you've checked. Well, I guess there were, right? So any problem where, you know,

01:14:36.080 | wherever there's a solution, there is a short witness that can be easily like a polynomial-size

01:14:42.320 | witness that can be checked in polynomial time. That we call an NP problem, okay?

01:14:48.320 | >>Beautiful.

01:14:48.800 | >>And yeah, so every problem that's in P is also in NP, right? Because, you know,

01:14:55.120 | you could always just ignore the witness and just, you know, if a problem is in P,

01:14:58.800 | you can just solve it yourself. Okay? But now the, in some sense, the central, you know,

01:15:04.640 | mystery of theoretical computer science is, is every NP problem in P. So if you can easily check

01:15:11.760 | the answer to a computational problem, does that mean that you can also easily find the answer?

01:15:17.840 | >>Even though there's all these problems that appear to be very difficult to find the answer,

01:15:23.600 | it's still an open question whether a good answer exists. So what's your—

01:15:27.520 | >>Because no one has proven that there's no way to do it, right?

01:15:29.680 | >>It's arguably the most, I don't know, the most famous, the most maybe interesting,

01:15:36.560 | maybe you disagree with that, problem in theoretical computer science. So what's your—

01:15:39.920 | >>The most famous, for sure.

01:15:41.120 | >>P equals NP.

01:15:42.160 | >>Yeah.

01:15:42.640 | >>If you were to bet all your money, where do you put your money?

01:15:45.280 | >>That's an easy one. P is not equal to NP.

01:15:47.280 | >>Okay, so—

01:15:47.680 | >>I like to say that if we were physicists, we would have just declared that to be a law of nature.

01:15:52.480 | You know, just like, just like thermodynamics or something.

01:15:54.560 | >>That's hilarious.

01:15:55.280 | >>Just giving ourselves Nobel prizes for its discovery.

01:15:58.240 | Yeah, yeah, no, and look, if later it turned out that we were wrong, we just give ourselves—

01:16:02.640 | >>Another Nobel prize.

01:16:03.520 | >>More Nobel prizes, yeah. I mean, you know, but yeah, because we're—

01:16:08.000 | >>So harsh, but so true.

01:16:09.200 | >>I mean, no, I mean, I mean, it's really just because we are mathematicians or descended

01:16:14.720 | from mathematicians, you know, we have to call things conjectures that other people

01:16:19.760 | would just call empirical facts or discoveries, right? But one shouldn't read more into that

01:16:24.960 | difference in language, you know, about the underlying truth.

01:16:28.480 | >>So, okay, so you're a good investor and good spender of money. So then let me ask—

01:16:32.240 | >>I don't know that.

01:16:32.880 | >>Let me ask another way. Is it possible at all? And what would that look like if P indeed equals NP?

01:16:41.120 | >>Well, I do think that it's possible. I mean, in fact, you know, when people really

01:16:45.040 | pressed me on my blog for what odds would I put, I put, you know, two or three percent odds.

01:16:50.000 | >>Wow, that's pretty good.

01:16:51.280 | >>That P equals NP. Yeah, just be—well, because, you know, when P—I mean, you really have to

01:16:56.960 | think about, like, if there were 50, you know, mysteries like P versus NP, and if I made a guess

01:17:03.840 | about every single one of them, would I expect to be right 50 times, right? And the truthful

01:17:09.040 | answer is no, okay?

01:17:10.560 | >>Yeah.

01:17:11.040 | >>So, you know, and that's what you really mean in saying that, you know, you have,

01:17:15.760 | you know, better than 98% odds for something, okay? But so, yeah, you know, I mean, there could

01:17:22.640 | certainly be surprises. And look, if P equals NP, well, then there would be the further question

01:17:28.720 | of, you know, is the algorithm actually efficient in practice, right? I mean, Don Knuth, who I know

01:17:35.280 | that you've interviewed as well, right, he likes to conjecture that P equals NP, but that the

01:17:41.440 | algorithm is so inefficient that it doesn't matter anyway, right? Now, I don't know, I've listened

01:17:47.040 | to him say that. I don't know whether he says that just because he has an actual reason for

01:17:51.920 | thinking it's true or just because it sounds cool, okay? But, you know, that's a logical

01:17:58.000 | possibility, right, that the algorithm could be N to the 10,000 time, or it could even just be N

01:18:03.840 | squared time, but with a leading constant of—it could be a Google times N squared, or something

01:18:09.440 | like that. And in that case, the fact that P equals NP, well, it would, you know, ravage the

01:18:16.160 | whole theory of complexity, and we would have to, you know, rebuild from the ground up. But in

01:18:21.200 | practical terms, it might mean very little, right, if the algorithm was too inefficient to run.

01:18:27.360 | If the algorithm could actually be run in practice, like if it had small enough constants,

01:18:33.600 | you know, or if you could improve it to where it had small enough constants that it was

01:18:38.960 | efficient in practice, then that would change the world, okay?

01:18:42.400 | You think it would have, like, what kind of impact would it have?

01:18:44.320 | Well, okay, I mean, here's an example. I mean, you could—well, okay, just for starters,

01:18:49.600 | you could break basically all of the encryption that people use to protect the internet.

01:18:53.440 | That's just for starters.

01:18:54.240 | You could break Bitcoin and every other cryptocurrency, or, you know,

01:18:58.000 | mine as much Bitcoin as you wanted, right? You know, become a super-duper billionaire,

01:19:06.480 | right? And then plot your next move, okay?

01:19:09.360 | Right, that's just for starters. That's a good point.

01:19:11.280 | Now, your next move might be something like, you know, you now have, like, a theoretically optimal

01:19:17.440 | way to train any neural network, to find parameters for any neural network, right?

01:19:22.240 | So you could now say, like, is there any small neural network that generates the entire content

01:19:27.840 | of Wikipedia, right? If, you know—and now the question is not, can you find it? The question

01:19:33.680 | has been reduced to, does that exist or not? If it does exist, then the answer would be, yes,

01:19:39.360 | you can find it, okay, if you had this algorithm in your hands, okay? You could ask your computer,

01:19:46.320 | you know, I mean, P versus NP is one of these seven problems that carries this million-dollar

01:19:51.360 | prize from the Clay Foundation. You know, if you solve it, you know, and others are the Riemann

01:19:56.560 | hypothesis, the Poincare conjecture, which was solved, although the solver turned down the prize,

01:20:03.120 | right, and four others. But what I like to say, the way that we can see that P versus NP is the

01:20:09.040 | biggest of all of these questions, is that if you had this fast algorithm, then you could solve all

01:20:14.720 | seven of them, okay? You just ask your computer, you know, is there a short proof of the Riemann

01:20:19.680 | hypothesis, right? You know, that a machine could—in a language where a machine could verify

01:20:24.640 | it, and provided that such a proof exists, then your computer finds it in a short amount of time,

01:20:30.080 | without having to do a brute force search. Okay, so I mean, those are the stakes of what we're

01:20:34.480 | talking about. But I hope that also helps to give your listeners some intuition of why I and most of

01:20:42.240 | my colleagues would put our money on P not equaling NP. —Is it possible—I apologize, this is a really

01:20:48.720 | dumb question, but is it possible to—that a proof will come out that P equals NP, but an algorithm

01:20:57.280 | that makes P equals NP is impossible to find? Is that like crazy? —Okay, well, if P equals NP,

01:21:04.880 | it would mean that there is such an algorithm. —That it exists, yeah. —But, you know, it would

01:21:12.880 | mean that it exists. Now, you know, in practice, normally the way that we would prove anything

01:21:18.240 | like that would be by finding the algorithm. —By finding one algorithm. —But there is such a thing

01:21:22.720 | as a non-constructive proof that an algorithm exists. You know, this has really only reared

01:21:28.000 | its head, I think, a few times in the history of our field, right? But, you know, it is theoretically

01:21:34.560 | possible that such a thing could happen. But, you know, there are—even here, there are some amusing

01:21:40.960 | observations that one could make. So there is this famous observation of Leonid Levin, who was, you

01:21:47.360 | know, one of the original discoverers of NP completeness, right? And he said, "Well, consider

01:21:52.400 | the following algorithm that, like, I guarantee will solve the NP problems efficiently, just as

01:21:59.520 | provided that P equals NP." Okay? Here is what it does. It just runs, you know, it enumerates every

01:22:06.640 | possible algorithm in a gigantic infinite list, right? From like, in like alphabetical order,

01:22:12.720 | right? You know, and many of them maybe won't even compile, so we just ignore those, okay? But now

01:22:18.080 | we just, you know, run the first algorithm, then we run the second algorithm, we run the first one

01:22:23.440 | a little bit more, then we run the first three algorithms for a while, we run the first four

01:22:28.080 | for a while. This is called dovetailing, by the way. This is a known trick in theoretical computer

01:22:34.960 | science, okay? But we do it in such a way that, you know, whatever is the algorithm out there in

01:22:41.760 | our list that solves NP complete, you know, the NP problems efficiently, will eventually hit that

01:22:48.000 | one, right? And now the key is that whenever we hit that one, you know, by assumption it has to

01:22:54.560 | solve the problem, it has to find the solution, and once it claims to find the solution, then we

01:22:59.680 | can check that ourself, right? Because these are NP problems, then we can check it. Now, this is

01:23:05.360 | utterly impractical, right? You know, you'd have to do this enormous, exhaustive search among all

01:23:11.680 | the algorithms, but from a certain theoretical standpoint, that is merely a constant pre-factor.

01:23:17.680 | >> That's at least possible.

01:23:18.800 | >> That's merely a multiplier of your running time. So there are tricks like that one can do to say

01:23:23.600 | that in some sense the algorithm would have to be constructive. But, you know, in the human sense,

01:23:30.480 | you know, it is possible that, you know, it's conceivable that one could prove such a thing

01:23:35.280 | via a non-constructive method. Is that likely? I don't think so, not personally.

01:23:41.360 | >> So that's P and NP, but the Complexity Zoo is full of wonderful creatures.

01:23:46.720 | >> Well, it's got about 500 of them.

01:23:48.480 | >> 500. So how do you get...

01:23:50.480 | >> More?

01:23:52.640 | >> Yeah, how do you get more? How are beings made?

01:23:56.160 | >> I mean, just for starters, there is everything that we could do with a conventional computer

01:24:02.560 | with a polynomial amount of memory, okay? But possibly an exponential amount of time

01:24:08.080 | because we get to reuse the same memory over and over again. Okay, that is called PSPACE,

01:24:13.360 | okay? And that's actually, we think, an even larger class than NP. Okay, well,

01:24:20.240 | P is contained in NP, which is contained in PSPACE. And we think that those containments

01:24:25.680 | are strict.

01:24:26.240 | >> And the constraint there is on the memory. The memory has to grow

01:24:30.240 | polynomially with the size of the process.

01:24:32.880 | >> That's right, that's right. But in PSPACE, we now have interesting things that were not in NP,

01:24:38.240 | like as a famous example, you know, from a given position in chess, you know, does white or black

01:24:45.200 | have the win? Let's say, assuming, provided that the game lasts only for a reasonable number of

01:24:50.720 | moves, okay? Or likewise for Go, okay? And, you know, even for the generalizations of these games

01:24:57.040 | to arbitrary size boards, right? Because with an eight by eight board, you could say that's just a

01:25:01.360 | constant size problem. You just, you know, in principle, you just solve it in O of one time,

01:25:06.240 | right? But so we really mean the generalizations of, you know, games to arbitrary size boards here.

01:25:14.080 | Or another thing in PSPACE would be like, I give you some really hard constraint satisfaction

01:25:21.920 | problem, like, you know, a traveling salesperson or, you know, packing boxes into the trunk of

01:25:28.880 | your car or something like that. And I ask not just, is there a solution, which would be an NP

01:25:33.920 | problem, but I ask how many solutions are there, okay? That, you know, count the number of valid

01:25:41.200 | solutions. That actually gives those problems lie in a complexity class called sharp P, or like,

01:25:49.120 | it looks like hashtag, like hashtag P. >> Got it.

01:25:51.760 | >> Okay, which sits between NP and PSPACE. There's all the problems that you can do in

01:25:57.920 | exponential time, okay? That's called EXP. So, and by the way, it was proven in the 60s that EXP

01:26:08.400 | is larger than P, okay? So we know that much. We know that there are problems that are solvable in

01:26:14.400 | exponential time that are not solvable in polynomial time, okay? In fact, we even know,

01:26:20.560 | we know that there are problems that are solvable in N cubed time that are not solvable in N squared

01:26:25.920 | time. >> And that, those don't help us with the controversy between P and NP at all.

01:26:29.840 | >> Unfortunately, it seems not, or certainly not yet, right? The techniques that we use to

01:26:36.400 | establish those things, they're very, very related to how Turing proved the unsolvability of the

01:26:41.120 | halting problem, but they seem to break down when we're comparing two different resources,

01:26:46.640 | like time versus space, or like, you know, P versus NP, okay? But, you know, I mean, there's

01:26:53.360 | what you can do with a randomized algorithm, right? That can sometimes, you know, has some

01:26:58.560 | probability of making a mistake. That's called BPP, bounded error probabilistic polynomial time.

01:27:05.200 | >> Wow. >> And then, of course, there's one that's very close to my own heart, what you can

01:27:09.360 | efficiently do, do in polynomial time using a quantum computer, okay? And that's called BQP,

01:27:16.000 | right? And so, you know, what-- >> What's understood about that class,

01:27:19.840 | maybe, as a comment. >> Okay, so P is contained in BPP, which is contained in BQP, which is contained

01:27:26.320 | in P space, okay? So anything you can, in fact, in like, in something very similar to sharp P.

01:27:33.280 | BQP is basically, you know, well, it's contained in like, P with the magic power to solve sharp P

01:27:39.760 | problems, okay? >> Why is BQP contained in P space?

01:27:44.480 | >> Oh, that's an excellent question. So there is, well, I mean, one has to prove that, okay? But

01:27:52.400 | the proof, you could think of it as using Richard Feynman's picture of quantum mechanics,

01:28:00.960 | which is that you can always, you know, we haven't really talked about quantum mechanics in this

01:28:06.640 | conversation. We did in our previous one. >> Yeah, we did last time.

01:28:09.600 | >> But yeah, we did last time, okay? But basically, you could always think of a quantum computation

01:28:16.160 | as like a branching tree of possibilities, where each possible path that you could take

01:28:24.000 | through, you know, the space has a complex number attached to it called an amplitude, okay? And now,

01:28:30.960 | the rule is, you know, when you make a measurement at the end, well, you see a random answer,

01:28:36.080 | okay? But quantum mechanics is all about calculating the probability that you're

01:28:40.720 | going to see one potential answer versus another one, right? And the rule for calculating the

01:28:47.120 | probability that you'll see some answer is that you have to add up the amplitudes for all of the

01:28:53.120 | paths that could have led to that answer. And then, you know, that's a complex number, so that,

01:28:58.560 | you know, how could that be a probability? Then you take the squared absolute value of the result.

01:29:04.400 | That gives you a number between zero and one, okay? So, yeah, I just summarized quantum mechanics in

01:29:10.960 | like 30 seconds, okay? >> Yeah, in a few sentences.

01:29:13.040 | >> But now, you know, what this already tells us is that anything I can do with a quantum computer,

01:29:19.440 | I could simulate with a classical computer if I only have exponentially more time, okay? And why

01:29:25.520 | is that? Because if I have exponential time, I could just write down this entire branching tree

01:29:31.920 | and just explicitly calculate each of these amplitudes, right? You know, that will be very

01:29:37.600 | inefficient, but it will work, right? It's enough to show that quantum computers could not solve the

01:29:43.600 | halting problem, or, you know, they could never do anything that is literally uncomputable in

01:29:48.960 | Turing's sense, okay? But now, as I said, there is even a stronger result, which says that BQP

01:29:55.520 | is contained in P space. The way that we prove that is that we say, "If all I want is to calculate

01:30:03.200 | the probability of some particular output happening, you know, which is all I need to

01:30:08.240 | simulate a quantum computer, really, then I don't need to write down the entire quantum state,

01:30:13.520 | which is an exponentially large object. All I need to do is just calculate what is the amplitude

01:30:20.240 | for that final state, and to do that, I just have to sum up all the amplitudes that lead to that

01:30:27.120 | state. Okay, so that's an exponentially large sum, but I can calculate it just reusing the same

01:30:33.280 | memory over and over for each term in the sum. >> And hence the P, in the P space.

01:30:37.760 | >> In the P space, yeah. >> Yeah.

01:30:39.360 | >> So what, out of that whole complexity zoo, and it could be BQP, what do you find is the most,

01:30:46.000 | the class that captured your heart the most? Was the most beautiful class that's just, yeah.

01:30:53.440 | >> Well, I used as my email address bqpqpolly@gmail.com, just because bqp/qpolly,

01:31:03.680 | well, amazingly, no one had taken it. >> Amazing.

01:31:06.720 | >> But this is a class that I was involved in defining, proving the first theorems about

01:31:14.240 | in 2003 or so, so it was close to my heart. But this is like, if we extended bqp, which is the

01:31:22.560 | class of everything we can do efficiently with a quantum computer, to allow quantum advice,

01:31:28.880 | which means imagine that you had some special initial state that could somehow help you do

01:31:35.440 | computation, and maybe such a state would be exponentially hard to prepare, but maybe somehow

01:31:44.080 | these states were formed in the Big Bang or something, and they've just been sitting around

01:31:48.000 | ever since, right? If you found one, and if this state could be like ultra-power, there are no

01:31:53.600 | limits on how powerful it could be, except that this state doesn't know in advance which input

01:31:59.760 | you've got, right? It only knows the size of your input. That's bqp/qpolly. So that's one that I

01:32:07.360 | just personally happen to love, okay? But if you're asking, there's a class that I think is way

01:32:16.800 | more beautiful or fundamental than a lot of people, even within this field, realize that it is.

01:32:24.000 | That class is called SZK, or statistical zero knowledge. And there's a very, very easy way to

01:32:31.760 | define this class, which is to say, suppose that I have two algorithms that each sample from

01:32:38.240 | probability distributions, right? So each one just outputs random samples according to possibly

01:32:45.360 | different distributions. And now the question I ask is, let's say distributions over strings of

01:32:51.520 | n bits, so over an exponentially large space. Now I ask, are these two distributions close or far

01:32:59.120 | as probability distributions? Okay, any problem that can be reduced to that, that can be put into

01:33:05.280 | that form, is an SZK problem. And the way that this class was originally discovered was completely

01:33:12.320 | different from that, and was kind of more complicated. It was discovered as the class

01:33:17.360 | of all of the problems that have a certain kind of what's called zero knowledge proof.

01:33:22.160 | The zero knowledge proofs are one of the central ideas in cryptography. Shafi Goldwasser and Sylvio

01:33:29.760 | Macaulay won the Turing Award for inventing them, and they're at the core of even some

01:33:35.200 | cryptocurrencies that people use nowadays. But zero knowledge proofs are ways of proving to

01:33:44.400 | someone that something is true, like that there is a solution to this optimization problem,

01:33:52.880 | or that these two graphs are isomorphic to each other or something, but without revealing why

01:33:58.640 | it's true, without revealing anything about why it's true. SZK is all of the problems for which

01:34:06.480 | there is such a proof that doesn't rely on any cryptography. And if you wonder, how could such

01:34:13.680 | a thing possibly exist? Well, imagine that I had two graphs, and I wanted to convince you

01:34:21.040 | that these two graphs are not isomorphic, meaning I cannot permute one of them so that it's the same

01:34:27.040 | as the other one. That might be a very hard statement to prove. You might have to do a very

01:34:33.760 | exhaustive enumeration of all the different permutations before you were convinced that

01:34:38.640 | it was true. But what if there were some all-knowing wizard that said to you, "Look,

01:34:43.760 | I'll tell you what. Just pick one of the graphs randomly, then randomly permute it,

01:34:49.120 | then send it to me, and I will tell you which graph you started with.

01:34:54.480 | And I will do that every single time." Right? >>Karl: Let me load that in. Okay,

01:34:59.280 | that's fine. I got it. >>Corey: Yeah. And let's say that that wizard did that 100 times, and it

01:35:04.080 | was right every time. Now, if the graphs were isomorphic, then it would have been flipping

01:35:09.600 | a coin each time. It would have had only a 1 in 2 to the 100 power chance of guessing right each time.

01:35:16.960 | So if it's right every time, then now you're statistically convinced that these graphs are

01:35:22.880 | not isomorphic, even though you've learned nothing new about why they are. >>Karl: So

01:35:27.440 | fascinating. >>Corey: So yeah. So SDK is all of the problems that have protocols like that one,

01:35:33.440 | but it has this beautiful other characterization. It's shown up again and again in my own work,

01:35:39.360 | in a lot of people's work. And I think that it really is one of the most fundamental classes.

01:35:44.720 | It's just that people didn't realize that when it was first discovered.

01:35:47.520 | >>Luis: So we're living in the middle of a pandemic currently. How has your life been

01:35:54.400 | changed? Or no, better to ask, how has your perspective of the world changed with this

01:35:59.600 | world-changing event of a pandemic overtaking the entire world?

01:36:03.680 | >>Karl: Yeah. Well, I mean, all of our lives have changed. I guess as with no other event since I

01:36:10.640 | was born, you would have to go back to World War II for something, I think, of this magnitude,

01:36:16.880 | on the way that we live our lives. As for how it has changed my worldview,

01:36:21.760 | I think that the failure of institutions like the CDC, like other institutions that we thought were

01:36:33.440 | trustworthy, like a lot of the media, was staggering, was absolutely breathtaking.

01:36:40.800 | It is something that I would not have predicted. I think I wrote on my blog that it's fascinating

01:36:50.000 | to re-watch the movie Contagion from a decade ago that correctly foresaw so many aspects of what was

01:36:58.800 | going on. An airborne virus originates in China, spreads to much of the world, shuts everything

01:37:07.680 | down until a vaccine can be developed, everyone has to stay at home. It gets an enormous number

01:37:16.240 | of things right. But the one thing that they could not imagine is that in this movie, everyone from

01:37:22.960 | the government is hyper-competent, hyper-dedicated to the public good.

01:37:28.560 | >>ZACH: The best of the best.

01:37:30.160 | >>KARL: Yeah, they're the best of the best. And there are these conspiracy theorists

01:37:36.480 | who think this is all fake news, there's not really a pandemic. And those are some random

01:37:42.960 | people on the internet who the hyper-competent government people have to oppose. In trying to

01:37:49.920 | envision the worst thing that could happen, there was a failure of imagination. The movie makers

01:37:56.960 | did not imagine that the conspiracy theorists and the incompetence and the nutcases would have

01:38:03.920 | captured our institutions and be the ones actually running things.

01:38:07.840 | >>ANDREW: So you had a certain... I love competence in all walks of life. I get so

01:38:13.680 | much energy, I'm so excited by people who do an amazing job. And I, like you,

01:38:17.520 | or maybe you can clarify, but I had maybe not intuition, but I hope that government at its

01:38:23.200 | best could be ultra-competent. First of all, two questions. How do you explain the lack of

01:38:30.400 | confidence? And the other, maybe on the positive side, how can we build a more competent government?

01:38:36.160 | >>KARL: Well, there's an election in two months. I mean...

01:38:39.680 | >>ANDREW: But you have a faith that the election process...

01:38:41.680 | >>KARL: It's not going to fix everything, but I feel like there is a ship that is sinking,

01:38:47.920 | and you could at least stop the sinking. But I think that there are much, much deeper problems.

01:38:54.160 | I mean, I think that it is plausible to me that a lot of the failures with the CDC,

01:39:03.440 | with some of the other health agencies even predate Trump, predate the right-wing populism

01:39:11.440 | that has taken over much of the world now. And I think that it is very... I've actually been

01:39:23.760 | strongly in favor of rushing vaccines. I thought that we could have done human challenge trials,

01:39:34.320 | which were not done. We could have had volunteers to actually get vaccines, get exposed to COVID.

01:39:46.080 | >>ANDREW: So innovative ways of accelerating what we've done previously over a long amount of time.

01:39:50.320 | >>KARL: I thought that each month that a vaccine is closer is like trillions of dollars.

01:39:56.560 | >>ANDREW: Are you surprised how slow...

01:39:57.760 | >>KARL: And of course lives, at least hundreds of thousands of lives.

01:40:02.160 | >>ANDREW: Are you surprised that it's taken this long? We still don't have a plan. There's still

01:40:06.480 | not a feeling like anyone is actually doing anything in terms of alleviating any kind of plan.

01:40:13.200 | So there's a bunch of stuff. There's vaccine, but you could also do a testing infrastructure where

01:40:17.680 | everybody's tested nonstop with contact tracing, all that kind of stuff.

01:40:21.200 | >>KARL: Well, I mean, I'm as surprised as almost everyone else. I mean, this is a historic failure.

01:40:26.800 | It is one of the biggest failures in the 240 year history of the United States. And we should be

01:40:33.440 | crystal clear about that. And one thing that I think has been missing, even from the more

01:40:41.520 | competent side is the World War II mentality. The mentality of, "Let's just... If we can,

01:40:54.640 | by breaking a whole bunch of rules, get a vaccine in even half the amount of time as we thought,

01:41:01.680 | then let's just do that because we have to weigh all of the moral qualms we have about doing that

01:41:11.280 | against the moral qualms of not doing it."

01:41:13.600 | >>ANDREW: And one key little aspect to that that's deeply important to me,

01:41:17.360 | and we'll go into that topic next, is the World War II mentality wasn't just about breaking all

01:41:23.680 | the rules to get the job done. There was a togetherness to it. So if I were president right

01:41:30.400 | now, it seems quite elementary to unite the country because we're facing a crisis. It's easy

01:41:38.800 | to make the virus the enemy. And it's very surprising to me that the division has increased

01:41:45.680 | as opposed to decreased. That's heartbreaking.

01:41:48.640 | >>KARL: Yeah, well, look, I mean, it's been said by others that this is the first time

01:41:52.160 | in the country's history that we have a president who does not even pretend to want to unite the

01:41:58.080 | country. I mean, Lincoln, who fought a civil war, said he wanted to unite the country.

01:42:06.560 | And I do worry enormously about what happens if the results of this election are contested.

01:42:17.520 | Will there be violence as a result of that, and will we have a clear path of succession?

01:42:22.320 | And look, I mean, we're going to find out the answers to this in two months, and if none of

01:42:28.880 | that happens, maybe I'll look foolish. But I am willing to go on the record and say,

01:42:33.200 | "I am terrified about that."

01:42:34.560 | >>ANDREW: Yeah, I've been reading The Rise and Fall of the Third Reich.

01:42:37.680 | So if I can--this is like one little voice to put out there that I think November

01:42:46.400 | will be a really critical month for people to breathe and put love out there. Do not,

01:42:53.360 | you know, anger in that context. No matter who wins, no matter what is said,

01:42:59.120 | will destroy our country, may destroy our country, may destroy the world because of

01:43:04.560 | the power of the country. So it's really important to be patient, loving, empathetic.

01:43:09.520 | One of the things that troubles me is that even people on the left are unable to have

01:43:15.840 | a love and respect for people who voted for Trump. They can't imagine that there's good

01:43:20.560 | people that could vote for the opposite side.

01:43:22.560 | >>ANDREW: Oh, I know there are, because I know some of them, right? I mean, you know,

01:43:26.640 | it's still, you know, maybe it baffles me, but you know, I know such people.

01:43:31.440 | >>ANDREW: Let me ask you this. It's also heartbreaking to me on the topic of cancel

01:43:36.320 | culture. So in the machine learning community, I've seen it a little bit, that there's

01:43:41.840 | aggressive attacking of people who are trying to have a nuanced conversation about things.

01:43:47.120 | And it's troubling because it feels like nuanced conversation is the only way to talk about

01:43:55.360 | difficult topics. And when there's a thought police and speech police on any nuanced conversation

01:44:02.320 | that everybody has to, like in a animal farm chant that racism is bad and sexism is bad,

01:44:08.880 | which is things that everybody believes, and they can't possibly say anything nuanced,

01:44:14.480 | it feels like it goes against any kind of progress, from my kind of shallow perspective.

01:44:19.440 | But you've written a little bit about cancel culture. Do you have thoughts there?

01:44:23.440 | >>ANDREW: Well, look, I mean, to say that I am opposed to, you know, this trend of cancellations

01:44:31.920 | or of, you know, shouting people down rather than engaging them, that would be a massive

01:44:36.000 | understatement, right? And I feel like, you know, I have put my money where my mouth is,

01:44:41.680 | you know, not as much as some people have, but, you know, I've tried to do something. I mean,

01:44:46.800 | I have defended, you know, some unpopular people and unpopular, you know, ideas on my blog. I've,

01:44:54.480 | you know, tried to defend, you know, norms of open discourse of, you know,

01:45:02.160 | reasoning with our opponents, even when I've been shouted down for that on social media,

01:45:07.760 | you know, called a racist, called a sexist, all of those things. And which, by the way, I should

01:45:12.160 | say, you know, I would be perfectly happy to, you know, say, you know, if we had time to say,

01:45:17.040 | you know, 10,000 times, you know, my hatred of racism, of sexism, of homophobia, right?

01:45:25.600 | But what I don't want to do is to cede to some particular political faction, the right to define

01:45:33.600 | exactly what is meant by those terms, to say, "Well, then you have to agree with all of these

01:45:38.720 | other extremely contentious positions, or else you are a misogynist, or else you are a racist,

01:45:45.440 | right?" I say that, "Well, no, you know, don't, like, don't I, or, you know, don't people like

01:45:52.720 | me also get a say in the discussion about, you know, what is racism, about what is going to be

01:45:58.480 | the most effective to combat racism, right?" And, you know, this cancellation mentality, I think,

01:46:06.240 | is spectacularly ineffective at its own professed goal of, you know, combating racism and sexism.

01:46:13.200 | What's a positive way out? So I try to, I don't know if you see what I do on Twitter, but on

01:46:20.160 | Twitter, I mostly, in my whole, in my life, I've actually, it's who I am to the core, is like,

01:46:26.000 | I really focus on the positive, and I try to put love out there in the world. And still,

01:46:31.440 | I get attacked. And I look at that, and I wonder, like...

01:46:36.560 | Oh, you too? I didn't know.

01:46:38.240 | Like, I haven't actually said anything difficult and nuanced. You talk about somebody like

01:46:43.920 | Steven Pinker, who, I actually don't know the full range of things that he's attacked for,

01:46:51.040 | but he tries to say difficult, he tries to be thoughtful about difficult topics.

01:46:55.440 | He does.

01:46:55.920 | And obviously, he just gets slaughtered by...

01:47:00.000 | Well, I mean, yes, but it's also amazing how well Steve has withstood it. I mean,

01:47:06.480 | he just survived an attempt to cancel him just a couple of months ago, right?

01:47:10.960 | Psychologically, he survives it too, which worries me, because I don't think I can.

01:47:15.360 | Yeah, I've gotten to know Steve a bit. He is incredibly unperturbed by this stuff.

01:47:19.920 | And I admire that, and I envy it. I wish that I could be like that. I mean, my impulse when

01:47:26.240 | I'm getting attacked is I just want to engage every single anonymous person on Twitter and

01:47:32.560 | Reddit who is saying mean stuff about me. And I want to just say, "Well, look, can we just

01:47:36.960 | talk this over for an hour, and then you'll see that I'm not that bad."

01:47:41.360 | And sometimes that even works. The problem is then there's the 20,000 other ones, right?

01:47:46.640 | And that's not... But psychologically, does that wear on you?

01:47:50.640 | It does, it does. But yeah, I mean, in terms of what is the solution, I mean, I wish I knew,

01:47:56.080 | right? In a certain way, these problems are maybe harder than P versus NP, right? I mean,

01:48:04.240 | but I think that part of it has to be for... I think that there's a lot of sort of silent support

01:48:11.200 | for what I'll call the open discourse side, the reasonable enlightenment side. And I think that

01:48:18.080 | support has to become less silent, right? I think that a lot of people agree that a lot of these

01:48:27.280 | cancellations and attacks are ridiculous, but are just afraid to say so, right? Or else they'll get

01:48:34.080 | shouted down as well, right? That's just the standard witch hunt dynamic, which of course,

01:48:38.880 | this faction understands and exploits to its great advantage. But more people just said,

01:48:47.760 | "We're not going to stand for this, right? Guess what? We're against racism too, but

01:48:57.760 | what you're doing is ridiculous," right? And the hard part is it takes a lot of mental energy. It

01:49:04.000 | takes a lot of time. Even if you feel like you're not going to be canceled or you're staying on the

01:49:09.920 | safe side, it takes a lot of time to phrase things in exactly the right way and to respond to

01:49:18.320 | everything people say. But I think that the more people speak up from all political persuasions,

01:49:27.760 | from all walks of life, then the easier it is to move forward.

01:49:32.720 | Since we've been talking about love, can you—last time I talked to you about the meaning of life a

01:49:39.600 | little bit, but here has—it's a weird question to ask a computer scientist, but has love for other

01:49:47.120 | human beings, for things, for the world around you played an important role in your life?

01:49:54.000 | Have you—it's easy for a world-class computer scientist, you could even call yourself a

01:50:03.200 | physicist, everything to be lost in the books. Is the connection to other humans, love for other

01:50:09.440 | humans played an important role? I love my kids. I love my wife. I love my parents.

01:50:19.600 | I am probably not different from most people in loving their families and in that being very

01:50:28.320 | important in my life. Now, I should remind you that I am a theoretical computer scientist.

01:50:36.320 | If you're looking for deep insight about the nature of love, you're probably looking in the

01:50:40.640 | wrong place to ask me, but sure, it's been important.

01:50:45.920 | But is there something from a computer science perspective to be said about love?

01:50:50.160 | Is that even beyond into the realm of consciousness?

01:50:56.480 | There was this great cartoon, I think it was one of the classic XKCDs, where it shows a heart,

01:51:04.560 | and it's like squaring the heart, taking the Fourier transform of the heart, integrating the

01:51:10.720 | heart, each thing, and then it says, "My normal approach is useless here."

01:51:16.880 | I'm so glad I asked this question. I think there's no better way to end this, Scott.

01:51:23.440 | I hope we get a chance to talk again. This has been an amazing,

01:51:26.080 | cool experiment to do it outside. I'm really glad you made it out.

01:51:29.120 | Yeah, well, I appreciate it a lot. It's been a pleasure,

01:51:31.520 | and I'm glad you were able to come out to Austin.

01:51:34.160 | Thanks.

01:51:35.600 | Thanks for listening to this conversation with Scott Aronson, and thank you to our sponsors,

01:51:41.200 | 8sleep, SimpliSafe, ExpressVPN, and BetterHelp. Please check out these sponsors in the description

01:51:48.480 | to get a discount and to support this podcast. If you enjoy this thing, subscribe on YouTube,

01:51:54.960 | review it with 5 Stars on Apple Podcasts, follow on Spotify, support on Patreon,

01:51:59.680 | or connect with me on Twitter @LexFriedman.

01:52:03.600 | And now let me leave you with some words from Scott Aronson that I also gave to you in the

01:52:08.320 | introduction, which is, "If you always win, then you're probably doing something wrong."

01:52:14.560 | Thank you for listening and for putting up with the intro and outro in this strange room in the

01:52:21.200 | middle of nowhere, and I very much hope to see you next time in many more ways than one.

01:52:27.600 | Bye.

01:52:29.140 | you

01:52:30.520 | you

01:52:31.020 | you

01:52:31.520 | [BLANK_AUDIO]

Scott Aaronson: Computational Complexity and Consciousness | Lex Fridman Podcast #130

Chapters