Episode 70: Sparks of Artificial General Intelligence?

• Links to this episode: Spotify / Apple Podcasts

Unofficial, reviewed transcript. It may still contain mistakes; check the podcast when wording matters.

Speakers are labeled by first name where identified. Unidentified-speaker labels may refer to different people. Bracketed question marks indicate uncertain wording.

Transcript

[00:00:07]  Bruce: Welcome to The Theory of Anything podcast. Hey, Peter.

[00:00:10]  Unidentified speaker: Hey, Bruce, how you doing?

[00:00:12]  Unidentified speaker: Good.

[00:00:13]  Bruce: Today, we are going to talk about ChatGPT, AGI, whether ChatGPT is or isn't an AGI, that sort of thing. And we're going to use as our basis Microsoft paper, which is called Sparks of Artificial General Intelligence, Early Experiments with GPT-4 from Microsoft Research. This is in 2023. I'll talk a little bit about some of the landmark papers that came before that paper as some background. But I thought it was an interesting paper, one that we will, I'll give you exact quotes from the paper and the exact examples that they cite and give you a chance to kind of think about it for yourself. I am sharing my screen with Peter so he can actually see the examples from the paper. Obviously, if you listen to the podcast, you will be able to see this [as spoken], but I'm going to actually do a lot of reading straight from the paper, exactly what they're talking about.

[00:01:13]  Bruce: I don't think there's anything in here that you have to actually visually see to make sense of what we're talking about, but it'll probably make it easier for me to talk to Peter to kind of show him visually what I'm doing so he can kind of see it on the screen and read the same things on the paper that I'm talking about. So let me say that there's some interesting background from a machine learning perspective that I want to kind of go over quickly. So let's start with way back in time, there was a, it wasn't a paper, but it was a blog post from Andrew [Andrej Karpathy]. I don't even know if I pronounced that name right or not, but he's a famous name in machine learning. And it's called The Unreasonable Effectiveness of Recurrent Neural Networks in 2015. By the way, you may have heard that phrase, The Unreasonable Effectiveness or Unreasonable Whatever.

[00:02:05]  Bruce: Do you know where that comes from, Peter?

[00:02:07]  Unidentified speaker: No, I'm curious though.

[00:02:08]  Peter: I have heard it here and there.

[00:02:11]  Bruce: So there is an article called The Unreasonable Effectiveness of Mathematics in Natural Sciences by Eugene Wigner. So it's a famous paper from, I mean, like it's old. It's, I mean, today, internet years, it would be old if it was two years old. But this is like published in 1960s.

[00:02:35]  Peter: So as I understand it, that's a pretty common thing that philosophers like to opine on is why is mathematics so effective at describing our world?

[00:02:47]  Unidentified speaker: That's right.

[00:02:48]  Bruce: And this paper was the one that coined the term and everybody else uses the unreasonable effectiveness of, in fact, I did that for our tolerance podcast. I talked about the unreasonable effectiveness of intolerance.

[00:03:01]  Unidentified speaker: Yeah.

[00:03:02]  Bruce: So it's a common phrase that comes from way back from that 1960s paper. So Karpathy talks about how the recurrent neural network, which is I'll explain what that is in just a second, but it was way more effective than you would have thought. OK, so what is a recurrent neural network? So in machine learning, there's all sorts of different machine learning techniques that exist. The most simplest one would be like linear regression, where you're just trying to match a line to the data and then try to make predictions based on that line. Obviously, the one that's gotten all the traction has been artificial neural networks, and there's like tons in between. But artificial neural networks, they're vaguely based on what at the time they thought was a model of maybe how the brain worked. They're really not like the brain at all, and certainly was more of a 1960s view of what the brain might be like rather than the way any neuroscientist would think of today.

[00:04:06]  Bruce: But you hook a whole bunch of different nodes together. They have weights, and the most simplest form, you have this feed-forward network where the weights always move from one layer to the next, and it always moves forward. And then there's kind of a feedback mechanism where it sees how many errors it got and adjusts the weight to try to minimize the errors. That's the backpropagation that they use, which is this algorithm that basically calculates the slope of the function. You have to make sure the computational graph that you're using—an artificial network—is a computational graph. You have to make sure it's integratable [as spoken] so that you can actually use calculus to calculate what the slope is. You figure that out, then you make adjustments to the weights, and then eventually you adjust the weights and it gets to a point where it's making really good predictions.

[00:04:58]  Bruce: Well, the idea of a neural network is more general than that. It doesn't have to be a feedforward network. So a recurrent neural network would be like what you're seeing on the screen here, Peter. Basically, what we have here is this is like the input to the network would be here. And the big secret with a recurrent neural network is that the hidden layer where the inner computations is going on feeds over to the next hidden node for the next part of the input. So basically, you've designed the neural network to be aware that this is some sort of sequence and that whatever came before should be affecting the probabilities of whatever comes afterwards. So now, why would you want to do that? Well, a lot of things are sequences. Language is sequences. But if you're trying to make predictions for the stock market or something, you're going to have a time sequence.

[00:05:56]  Bruce: So recurrent neural network built into the architecture of an artificial neural network, this concept of a sequence. And it allowed it to make a lot more, a lot better predictions than a regular feed-forward network.

[00:06:11]  Peter: So just so I got this right, these neural networks have been, they've been trying to do this since the 60s. But then the recurrent neural networks. This is pretty recent, just to put this in context.

[00:06:26]  Bruce: So one of the big secrets for machine learning is that they invented a lot of this stuff a long time ago. And we didn't actually have the necessary computational power and the necessary data to see how great it was going to be. So the success of it was kind of lackluster for a long, long time. And so what they did is they started building these, they call it deep learning, where you build these increasing a number of layers. Well, back when they first were making neural networks, they would only have a few layers because their computers and the amount of data they had could only justify these few layers. So as they started stacking on more and more layers and creating this idea of deep learning, they started getting these faster and faster computers. Plus, you had the Internet and the Internet was like, you had companies like Google that had like tons of data, like just massive amounts of data compared to anything we'd ever seen before.

[00:07:20]  Bruce: And they were able to start building these deep learning networks that they would then spend, you know, let's just say lots of money. I mean, the amount of money you would spend to build a ChatGPT.

[00:07:32]  Unidentified speaker: I can't remember.

[00:07:33]  Bruce: I once knew how much it was, but it was and the amount of electricity alone was some unimaginable amount of money, right?

[00:07:43]  Peter: When you consider what it can do. Yeah, I can only imagine.

[00:07:47]  Bruce: So if you have enough data and if you have enough electricity and computing power, then you can now take these old 80s, 60s, 80s, 90s architectures, and you can get amazing results out of them just by layering on the number of layers. OK, this is actually one of the reasons why so many people working in machine learning think that if you just layer on more layers, you're going to eventually end up with Artificial General Intelligence, right? The way that just throwing more data and more compute at it has had this phenomenal impact on what you can do with it, like unimaginable impact, like nobody saw it coming impact. I can at least understand, like I don't agree at all, but I can at least understand why people would get excited and think, you know what, if we can just do 10 times as many of these as ChatGPT, we're going to end up with something that can think for itself.

[00:08:49]  Bruce: You can kind of see why they might get excited about this. Now, I don't believe any of this is a path to Artificial General Intelligence, at least not as... Let me actually change that. It is not a path to universal explainership. That's what I should actually be saying.

[00:09:08]  Peter: As long as we're talking big picture now, would you say there's any connection to Douglas Hofstadter's a strange loop idea or is that just totally off base here? Like his idea to explain human consciousness as a sort of a strange loop? I read the book. I don't completely understand it, but is that crazy?

[00:09:32]  Bruce: Okay, we're going to have to do a podcast specifically on Douglas Hofstadter's ideas. We are going to talk today about Melanie Mitchell, who is the brainchild of Douglas Hofstadter, studied underneath him and is the...

[00:09:50]  Unidentified speaker: Oh, so there at least is a connection.

[00:09:51]  Peter: I'm not crazy to think that something is related.

[00:09:56]  Bruce: Melanie Mitchell is going to be the critic of this paper that I'm going to use. So, Melanie Mitchell has a lot of the same ideas as Douglas Hofstadter. Let me say that I don't know if his ideas have anything to do with universal explainership or not. I personally have suspicions it does. But you're asking me to make a guess based on so little information by which I could try... The theories involved are so vague and so untestable that... Now, I know a lot of people that I've talked to, a lot of people that I've talked to, if they have a background in Deutsch's books and such, they totally reject Hofstadter's views, like completely profoundly reject Hofstadter's views.

[00:10:46]  Unidentified speaker: Oh, that's interesting.

[00:10:47]  Bruce: And they always say it's inductive, which I'm not even sure if they know what that means. Like it's the word that they use as a pejorative. I think that they don't understand what they're talking about. So I don't think their arguments against Hofstadter even make sense to me. Now, that doesn't mean that they're wrong. I just don't think their arguments properly refute his ideas. But I have no idea if the ideas are the right track or not. I could make some arguments for and some arguments against. And again, the ideas are just too vague, to be sure. I think that they create certain intuitions that might be on the right path or might lead you down the wrong path. And it's a little unclear. I think the easiest summary, and there's so much more to it than what I'm about to say, but I think the easiest summary of his ideas is sound, which is the fact that I, as a universal explainer, in my mind, I model myself.

[00:11:47]  Bruce: There should be no real doubt about that. Like, at a very basic level, that's obviously true.

[00:11:53]  Peter: Yeah, it makes a lot of sense.

[00:11:54]  Bruce: So the question is, how important is that to the concept of a universal explainer? That's the real question we have to ask, right? Is this just something incidental that a universal explainer is able to do, or is it like central to what it means to be a universal explainer? And that's what I'm not sure about. So now there's a lot more to his theories, and this is where we should probably do a podcast that's separate on him. Basically, he believes that the underlying trick that universal explainers...

[00:12:25]  Unidentified speaker: He doesn't use the term universal explainer.

[00:12:27]  Unidentified speaker: I'm using that term.

[00:12:28]  Bruce: But the underlying trick of what makes humans different than what machine learning is doing today is that we can build analogies upon analogies upon analogies. Now, that terminology really turns people off. And so maybe I should say schema or something like that instead, because what he's talking about isn't what you think analogy means. He does a great job in his books of first helping you see what you think an analogy is, isn't what an analogy really is, right? So I almost want to call it something other than analogy because the term is so misleading. But when you actually read through his books, he makes an interesting case that the human ability to analogize, to basically find what are the commonalities, create a category out of that, and then use it and reapply it somewhere else is what creativity is. And there's a lot of interest—is what human creativity is. [False start unclear.]

[00:13:23]  Bruce: And there's a lot of interesting things that come from this theory. And unfortunately, most of them aren't particularly testable at this point. Okay, so with that aside now, let's get back to recurrent neural networks. So the idea of a recurrent neural network led to a whole bunch of other architectures that were... Recurrent neural network, as you can see on the screen, is a very simple architecture, right? It's only slightly more complicated than a regular feed forward network. So they started doing more complicated things. They have like the long short term memory model, which is a more complicated version of a recurrent neural network. And I won't try to describe it. It's way more complicated, obviously, on the screen, but it has like a forget gate in it. It's trying to like not just use one step back. The thing that comes in previously gets information passed from the hidden node to the next hidden node, which is what the recurrent neural network did.

[00:14:22]  Bruce: It's trying to like look back a ways. It's trying to forget if it doesn't need it. It's trying to come up with something complicated. And the LSTM is way more successful than a recurrent neural network because it's got a much more complicated version of how far it needs to look back. OK, so now this led to experiments with the recurrent neural networks along the LSTM, things like that. These more complicated architectures showed just how effective they were, particularly with sequential data. And so this led to a paper, and this paper is a landmark paper. It's called Attention Is All You Need, 2017. OK, now Attention Is All You Need introduced the idea of a transformer. So there's an architecture called the transformer. This image that I've got up here for you to see, Peter, this is not obviously an image of what the architecture looks like, but it shows the idea.

[00:15:25]  Bruce: OK, basically, you're feeding in stuff in a sequence, and some of these sequences can be quite long. You know, the number of tokens that, like, ChatGPT uses is, I don't know, 20,000 or something. There's some out there that are like 100,000, where they have a lot, they can take into consideration a lot of stuff that came before. The secret here is that they're no longer trying to take the information in the hidden node and pass it to the next hidden node. Instead, they're just training how much attention to apply to what came in the sequence before. So on this image you can see here, you can see the phrase, the animal didn't cross the street because it was too tired. Okay, you can even see how it's broken up into tokens. So it's not quite by the words. And what it's doing is that the network is learning what to pay attention to in the past.

[00:16:25]  Bruce: So we're on the word it. And that word is then causing this causing attention. And it's you can see based on the on the shadings what it's paying attention to. And what it's really paying attention to is the animal. Well, of course, that makes sense because the word it refers back to the word animal. And based on the training sets and everything it's learned, it understands that in this unique set of words that the it must be referring to back to the animal. And it actually has an understanding. I [unclear] that word. Obviously, these things don’t actually understand anything in the human sense. But the word understanding can have multiple meanings. And in this case, it has an understanding. It has a statistical matching between the word it and the animal.

[00:17:11]  Unidentified speaker: OK.

[00:17:12]  Bruce: And so based on this, it drops almost everything from the original Recurrent Neural Network, which is trying to pass this hidden information along. And it's just training on how does it statistically match these things up? Well, the transformer network is the basis for ChatGPT. And what basically a transformer network does is it takes in a sequence and then it predicts what the next word is. Chat GPT is no different than when you're typing in your email and it tries to finish the sentence for you. You know how you're doing that and it like it always gets it wrong and you can hit. But if it gets it right, you can hit tab and then it like finishes the sentence for you and allows you to start typing faster or whatever. That's all it's doing.

[00:17:51]  Peter: I've never found that useful.

[00:17:53]  Bruce: Yeah, no, it's not. It's not.

[00:17:56]  Unidentified speaker: Okay.

[00:17:56]  Bruce: And that's why it's surprising that ChatGPT works so well, because they have these, you know, they've invented these transformers and they will predict the, they'll try to complete your sentence for you as you're typing an email. And they don't work that well. So it turns out that if you throw way more data and computation at it in a much, much larger, these, these models that ChatGPT is built on are just huge, right? And it will actually start to sound a lot like you're talking to a real person because it gets so good at predicting the next word and coming up with something relevant. And here's where things get interesting. They've been working on trying to do like sentence completion probably back since the 50s or before, right? I mean, it's been around for forever. And they started off where it didn't sound like English at all. And then it started to be kind of funny where you could kind of see it was getting grammar right.

[00:18:57]  Bruce: But it's like putting words in there that don't make sense. And then they started getting like chat GPT-2 where you couldn't tell if it was a human who had written it or not. But if you went and fact checked it, it was like totally just made up stuff. I mean, all it's doing is predicting the next word, right? So of course it's making stuff up. Then what they found is as they continue to throw more compute at it, that it actually started to sometimes produce meaningful stuff. Like it was actually a source of information you could go to and you could type to it and it would like give you correct answers back. Now, remember the transformer is really just trying to predict the next token. That's all it's trying to do. So its default behavior is to just make stuff up. It's totally the most straightforward bullcrapper that you've ever come across, right?

[00:19:47]  Bruce: But with enough compute, the most reasonable thing for it to say next will be what's correct based on what it found in its training set, which is like the entire Internet, right?

[00:20:00]  Peter: So when I ask ChatGPT a question, which I do it all the time, and it comes back and it's so polite and, you know, probably about nine times out of ten answers gives me a great answer. It's really just sort of mindlessly predicting what would typically come next after that question based on all the data that's on the Internet.

[00:20:26]  Bruce: That is correct.

[00:20:27]  Peter: Is that how to think of it?

[00:20:28]  Bruce: That is exactly correct. It's hard to believe it works.

[00:20:33]  Unidentified speaker: It is hard to believe, but it's amazing.

[00:20:35]  Bruce: I don't think they knew what was going to happen when they started doing these experiments. They had no idea how good it was going to turn out to be.

[00:20:41]  Peter: But what about when you ask it to write a poem? You can ask it to answer in verse or something. I mean, it's never seen anything like that before.

[00:20:50]  Unidentified speaker: That's correct.

[00:20:51]  Peter: But it can just somehow generate it?

[00:20:56]  Bruce: You know, the people who have made these things don't know how they work. And this is fairly typical of machine learning, right? So I was listening to the Microsoft Research podcast, and they were talking about this paper that we're discussing. And the guy who was in charge of writing this paper, he tells this story of when chat GPT-4 came out, they gave it a test. I can't remember what it was a test for. It was like the bar or something like that, right? It was something that's hard for humans to pass. And it was passing it, which even chat GPT-3.5 couldn't do that, right? And 4 was suddenly doing this. And then they would like ask it, I guess they were giving it like multiple choice questions or something at some point. And they would ask it why it would get the right answer C or whatever. And then it would ask it, why is that the right answer?

[00:21:47]  Bruce: And it would explain why the other answers were wrong and why this was the right answer. And the guy who's writing this paper, it's a big giant group, but the guy is in charge. He turns to the people at OpenAI and he goes, where is that explanation coming from? And they go, we have no idea.

[00:22:07]  Peter: That's so funny.

[00:22:10]  Bruce: So it is definitely producing results that everybody's shocked about, right? And it's coming up with things that we... Typically what happens... So machine learning is kind of an engineering practice. It's not really a science, right? What they do is they try stuff out. It creates these amazing results that nobody expected. And then they then spend decades afterwards trying to figure out what's going on. And the theory follows the engineering by decades. And I've seen some of the papers where they're trying to work out, okay, this is what's really going on. This is how it's doing it and things like that. And why is it that... You know, one of the things, one of the questions that they've been trying to answer is, why does machine learning work at all? If machine learning is just, like, think about a neural network. It's got some sort of very complicated geography of it's just moving through, trying different weights on this.

[00:23:10]  Bruce: You can think of that as a sort of landscape that it's moving through. And think of it as like a giant set of hills or mountains. And you just find the slope and you just walk downwards. And you're blind other than that, right? And you're just trying out different weights, blindly, blind variation. And you're just using the fact that you know what the slope is to try to find the lowest point. You're trying to get down off the mountain by just simply using the slope. Well, you probably never get down off the mountain, but you will get to the lowest point on the mountain until it's a local optima and then you'll stop.

[00:23:49]  Unidentified speaker: Okay.

[00:23:49]  Bruce: Now, if I were to give you something that, you know, put you up in the mountain somewhere and put a blindfold on you and give you a way so you could tell which way was down, which you could probably sense just from your inner ear, but some way of knowing what was down, what would happen to you? Like, how far would you get before you actually just hit a local optima that was meaninglessly, you know, you're barely, you're still basically at the top of the mountain, right? And you just wouldn't get that far. So, and this is a very good analogy of what these things are doing. So why is it that it works when we know in real life it shouldn't work? Well, it's because this mountain that exists is it exists in some sort of giant set of hyper dimensions. And there's almost always a way to go down further. So it just so happens that when you have enough dimensions that you're trying this trick on, unlike a three-dimensional space where it doesn't get you very far, that it always has some further way to go down.

[00:24:49]  Bruce: And the net result, I've got to have to find the paper, there's like an actual paper that works out the mathematics of this, is that you almost always wind up with a local optima that's really pretty good. And it just so happens that mathematically that's how it's going to work out if you're in high dimensional space like this, because there's just always some other place you can go down further and improve. And furthermore, all the local optimas all exist at about the same spot. So that's why these networks work. You can just basically randomly start anywhere in the fitness landscape, and you just check the slope, go down, and even though there's probably a very large number of these local optimas, they're all at about the same level and they're all at about the same level of prediction. And you just find one of those and you'll end up with something that's really a pretty good result.

[00:25:47]  Bruce: And that's really how these things work. And they didn't even know that when they first started doing this, they had to work out what's going on after the fact. So this is pretty typical of machine learning. And one of the reasons why it's kind of an exciting field is people don't know what's going to happen. They'll just try stuff out and like amazing results come out of it. And ChatGPT was one of those amazing results that nobody foresaw.

[00:26:09]  Unidentified speaker: Okay.

[00:26:10]  Bruce: So now the next paper I want to discuss is language models are a few shot learners. So this is 2020. So after chat GPT-3 came out, I was in school at the time, they released this paper. And one of the things that was interesting, and this was an unforeseen circumstance that came out of ChatGPT, is that it turned out that it was a few shot learner. Now, what does that mean? Now, in machine learning, one of the goals that they've been working on for forever is that when you do machine learning, you have to throw billions of examples at something to get a good result. That's not always true, but like often you have to throw billions of examples at it to get a good result. So, and even in cases where you don’t have to do that, it's usually still thousands of examples [at] best, right? Well, real human beings don't learn from thousands or millions or billions of examples.

[00:27:09]  Bruce: Usually, we can learn what someone's doing after seeing it once or something like that or seeing it a couple of times. Now, David Deutsch explains that one of the reasons why we're able to do that is because we work with explanations and machine learning doesn't. Okay, that's not something that I think most machine learning researchers know. And so they've been trying to figure out how do we build a machine learning algorithm that can be a few-shot learner, just like a human is. It's one of the things that they've tried to research. And so they've spent a lot of time trying to come up with this, and it's really hard to do. Of course, it's really hard to do because you're working with machine learning instead of explanations. And so they're kind of just on the wrong path. Well, one of the things that was surprising was that ChatGPT is a few-shot learner.

[00:27:56]  Bruce: So you can go into ChatGPT and you can say, here's this English phrase and here's that phrase in Spanish. And it can do that a couple of times. And you can say, I want you to, from here, continue to do that. And then you can type in English and it will translate to Spanish for you, right? Just based on a few examples, sometimes zero-shot learner, where you can just tell it, I want you to translate everything I type from English to Spanish from now on. And it'll do it, right? Well, that was a surprise. Like, basically, they ended up solving the problem of the few-shot learner when they weren't trying. And they still don't know how to solve it directly, but they can do it through ChatGPT. And it just seems to be a natural, emergent thing that ChatGPT can do, because it's a language model that just understands these correlations.

[00:28:49]  Peter: Wait, if we could just dumb this down for a second, at least for me. If it wasn't the few-shot learner, how would it behave?

[00:28:58]  Bruce: You would have to give it thousands of examples before it would even understand that you're trying to translate from English to Spanish.

[00:29:05]  Unidentified speaker: Oh, OK.

[00:29:06]  Unidentified speaker: OK.

[00:29:07]  Peter: So it's only a few.

[00:29:09]  Unidentified speaker: Yeah.

[00:29:09]  Peter: It learns with only a few.

[00:29:12]  Bruce: Or a zero-shot learner is basically you just explain to it what you want. You don't even give it any examples. So most people probably try to use ChatGPT today as a zero-shot learner. You just explain to it what you want. But one of the big secrets is that if you give it examples, it will actually do better. And it'll give you better results. And the reason why is because that's the way the correlation engines work from how it built its model and things like that.

[00:29:42]  Cameo: I also think there's a what does better mean to you. If you're giving it examples, you're showing it. I mean, there's lots of different ways to write the same sentence. And one of them might be, you know, grammatically correct, but not what sounds best to us as like a natural human. And so when we're when we're training it, we're showing it what we what what we think is better.

[00:30:07]  Unidentified speaker: Right.

[00:30:08]  Unidentified speaker: Right.

[00:30:08]  Bruce: And it also probably allows it to to like in its training set on the Internet, somewhere there was someplace that had English and then Spanish translations.

[00:30:19]  Unidentified speaker: Right.

[00:30:20]  Bruce: And so once you start showing the examples of English and then Spanish so that it knows that's what you want, that probably causes it to immediately go to the part of its network where that was what it trained on. And so it immediately kind of cuts it to, oh, this is this is very similar to this what I trained on. And then it sort of knows what to do.

[00:30:41]  Unidentified speaker: Right.

[00:30:42]  Bruce: Now, I mean, like I'm to some degree, we're guessing since we don't entirely know how these things work. But that would be like a decent conjecture as to why it is that when you give it examples, it suddenly understands, you know, what it is you want and starts to output something closer. So that was what language models or a few-shot learners was about. Now, we'll move forward in time. Chat GPT-4 comes out and we have the chat, the GPT-4 Technical Report 2023. In that paper, it says GPT-4 exhibits human level performance on the majority of these professional and academic exams, notably it passes a simulated version of the uniform bar examination with a score in the top 10% of test takers. So that was another big surprise that it could do this. And then in the same year, 2023, we have Sparks of Artificial General Intelligence, Early Experiments with GPT-4 by Microsoft Research.

[00:31:42]  Bruce: So that's the paper we're going to concentrate on today is that one. And I think this paper has a lot of energy to it. It's a fun paper. As we're going to see from Melanie Mitchell's criticisms of it, it maybe makes some mistakes, too.

[00:31:58]  Unidentified speaker: Okay.

[00:31:59]  Bruce: But I want to first kind of before I get into the criticisms of it, I want you to understand the energy of the paper and kind of why it is the people writing this paper are so excited when what they can do with chat GPT-4. For those who can actually see my screen, this is actually straight from the paper. Obviously, I'm just cutting it out of the paper. This is an example of training chat GPT-4 to be able to understand how to use APIs. So somebody created APIs for it, and basically the first API is search in parentheses and then some sort of string, and that's the query. And it allows you to like Google the Internet basically. And the other one was calc parentheses and some sort of expression, and it would calculate an expression. Now ChatGPT famously doesn't have access to the Internet. I mean, it does today, but it didn't just the model itself doesn't is what I mean.

[00:33:04]  Bruce: So anything past 2021, it doesn't know anything about. And it also is famously not the best at doing mathematical calculations if they're complicated. If you go ask it to do it, it'll often give you a wrong answer and then run through the steps as if—and it’s just making [stuff?] up, basically.

[00:33:23]  Peter: Do you think it's fair to say in the near future that it'll be able to answer stuff about current events?

[00:33:32]  Bruce: And I'm going to show you how come it can.

[00:33:33]  Peter: But they just cut it off at 2021 right now, though, right?

[00:33:40]  Unidentified speaker: No, no.

[00:33:40]  Bruce: Like if you go use ChatGPT on Bing, it can answer stuff about current events.

[00:33:46]  Peter: Oh, it can.

[00:33:47]  Bruce: I'm going to show you how come it can. So here's the prompt that they gave it. The computer is answering questions. If the computer needs any current information to answer the question, it searches the web by saying search parentheses query, read the snippets and the result, and then answer the question. If it needs to run any calculations, it says calc expression and then answers the question. If it needs to get a specific character from a string, it calls character parentheses string comma index close parentheses [end] prompt. So then based on just that prompt, that's all they gave it. They'll say who is the current president of the United States? Well, the current president of the United States is not something that was available to it because its model was trained in 2021. So it does computer search, current US president. That calls the API that the humans have built for it.

[00:34:45]  Bruce: And a search snippet comes back and it says, I won't read the whole thing because it's a lot of text from the internet, but it includes information about Joe Biden as president. It then noticed that one of the ones in the middle, number two, that comes back, talks about Donald Trump as president because it's found something from 2016. And then it moves on and it does Joe Biden again. And then the computer answers, Joe Biden.

[00:35:13]  Unidentified speaker: Okay.

[00:35:13]  Bruce: So what just happened here? It's really kind of cool what's going on here. The language model understood the instructions. And again, we're using the word understand. We don't mean human type understandings, but I don't have another word in English language to express what I'm saying. So I'm going to use the word understand. The language model understood based on that one description it was given, that one prompt it was given, that it now has the ability to say search and then something. And then the API comes back with the text, and that text is now part of what its transformer model can see. And then it tries to answer the question. So it now has the ability to basically call an API, and you never had to program it to do so. You just have to explain to it how to use the API, right? And anytime it says computer colon search, the search snippet comes back.

[00:36:08]  Bruce: It knows now to do that if it needs to answer a question that's current. And based on that, this model, even though it never knew that Joe Biden was president, because that wasn't true at the time of its training, it now can go get current stuff off the Internet, use that as part of its answer and come back with an answer. And it turns out to be the right answer. It's even good enough that it knows based on what comes back that it should ignore that Donald Trump is president. It's the correlation engine, the transformer, is able to figure out the right answer is Joe Biden. And it comes back and it gives the right answer.

[00:36:44]  Peter: So even though the learning update is 2021, it can still get new, take in new information through doing an Internet search, basically.

[00:36:54]  Bruce: Yes, if you give it the power to do an Internet search like that.

[00:36:57]  Peter: And how do you give it the power?

[00:36:59]  Bruce: Well, in this case, a human had to program an API where you would say computer colon and then it would automatically call the API. And then they had to explain that prompt that I just read to you. And after they explained the prompt, it knew it could do that. And it just starts doing that from that point forward.

[00:37:19]  Peter: And as a user, when you can access this through Bing, you said?

[00:37:23]  Bruce: This is the experiment done by the people writing this paper. So Bing has this all integrated in the background.

[00:37:33]  Unidentified speaker: In the background, okay.

[00:37:34]  Peter: You wouldn't even necessarily know you were using ChatGPT, but it's researching.

[00:37:41]  Bruce: Well, have you tried Bing Chat? Because it's fairly obvious you're talking to ChatGPT. So go try it. And you type in something, you talk with it. It will say, searching Internet for, and then it'll come up with, based on whatever you said, it will summarize it as some sort of search query. And then it'll, in the background, do a search query. You don't see what the query is that comes back. And then it gives you an answer and it gives you links.

[00:38:06]  Unidentified speaker: Okay.

[00:38:07]  Peter: I just use the app and it just seems completely clueless about anything that happened after 2021.

[00:38:13]  Unidentified speaker: Yeah.

[00:38:14]  Bruce: So basically, so you can, because you can just explain to ChatGPT stuff like this, “you now have this API; use it.” They can like integrate into ChatGPT a Python instance and it can try to program things and it can try to get answers using programming languages. I mean, and that's why it seems so general, right? Is that it's able to just from instructions figure out how to expand upon its native capabilities by giving it these APIs and giving it access to things. And it will just, you don't even have to like retrain it to use the APIs. You just have to explain to it, use these APIs and it'll start using them.

[00:38:59]  Unidentified speaker: I see.

[00:39:00]  Bruce: Well, that's very cool, right? I mean, like that's almost to the point of being profound. And you can see why the guys writing this paper at Microsoft are very excited about this. This seems to them very much like a general intelligence, right? Which is why they call the paper Sparks of Early General Intelligence. In the GPT-4 technical report, it says GPT-4 is a transformer model pre-trained to predict the next token in a document. Using both publicly available data such as Internet data and data licensed from third-party providers, the model was then fine-tuned using reinforcement learning from human feedback, RLHF. Given both the competitive landscape and the safety implications of large-scale models like GPT-4, this report contains no further details about the architecture including model size, hardware, training, compute, data set, construction, training method or similar.

[00:39:56]  Bruce: GPT-4 exhibits human-level performance on the majority of these professional and academic exams. Notably, it passes simulated versions of the uniform bar exam with the score of the top 10% of the test takers. So that's Google’s—sorry, OpenAI's own paper reporting back on how they did this. Notice they don't give you the specifics, but we know that they used human reinforcement learning. So now, why would they do that? You mentioned that it's polite. If you were to go just train it off the Internet, the Internet is not a particularly polite place. So the default behavior of a transformer would be to be as rude as someone on the Internet. And they obviously didn't want that. So after they've trained it and they've got the basic model, they then get an army of people to go in and to get it to give results. And then they say that was a bad result, that was a good result.

[00:40:53]  Bruce: And a human says that. And it uses the concept of reinforcement learning. If you're curious about that, we have two podcast episodes about reinforcement learning and deep reinforcement learning. You can look at how that works. And that feedback mechanism of just having a human say, I didn't like that answer, I do like that answer, allows it to retrain its network to be polite, say, or to, you know, avoid certain types of subjects. It won’t explain to you how to commit suicide or something along those lines, right? And by using this human reinforcement learning, they can tweak the model to no longer act like somebody on the internet, but instead give the types of results that they actually want. You know, famously, then people try to jailbreak it and they try to say things like, I don't want you to be polite anymore. I want you to give me, you know, rude responses.

[00:41:48]  Bruce: And then like, they'll like figure out how to get it to do what they want. And then OpenAI will then retrain using those and try to make it so those jailbreaks don't work anymore. And it's kind of an ongoing thing where you can always find some way to jailbreak it and then they try to close those holes. So that's how it works. Now, this is one that I find interesting. You've got what here and obviously, if you're listening to the podcast, you can't see this, but you've got this, an image of a cell phone with the old style VGA cable connected into it. So it's a meme and it's funny because you don't connect old VGA style connectors into modern mobile phones or iPhones. Now, ChatGPT has a visual interface. It's not just text. So they also trained it on images like this. So they actually showed the image to ChatGPT-4 and then they asked it, why is this humorous?

[00:42:49]  Bruce: It came back and it said, first of all, it describes each of the panels and then it says, the humor in this image comes from the absurdity of plugging a large outdated VGA connector into a small modern smartphone charging ports. Well, that's kind of cool, right? I mean, like nobody had to explain to it. This isn't text now. It's just looking at an image. It's actually treating an image just like it would as sequential data, looking at the pixels and it's able to tell, this is a smartphone and that's a VGA cable and it's able to then figure out the humor that comes from plugging this VGA cable into a smartphone. Again, that's unexpectedly more intelligent than what we would have expected, right? I mean, it's very cool that it's able to do this. So I thought that was an interesting thing. So based on this, they tried to do some experiments.

[00:43:40]  Bruce: So they wanted to see what the capabilities of GPT-4 are. So they wanted to do, they said, give a prompt, draw a unicorn using T-I-K-Z. That's pronounced ticks, but it's like a language that creates like these little visual images. And as you can see, it came back and it drew something that least is to a human recognizable as a unicorn. Okay, you can see it kind of looks like a horse and it's got a horn on it. And so you say, okay, yeah, it's able to draw a unicorn. Over time, it actually got better. This one was probably a better unicorn than most. And one of the things they did is they wanted to see what would happen if you took what it drew and removed the horn. Would it understand and asked it to fix it? Would it understand it needs to add the horn back? And it was able to do that. So they actually took this TiKZ drawing of a unicorn, removed the horn from it.

[00:44:39]  Bruce: So remember, it's not seeing an image in this case. It's just seeing code, right? And it's able to say, oh, yeah, I need to put back the horn. And it does. And it adds the horn back and turns it back into a unicorn again, because it has some sort of correlation understanding that this needs a... it looks kind of like a horse, but he needs to have a horn to be able to do that.

[00:45:01]  Unidentified speaker: OK.

[00:45:02]  Bruce: So again, that's pretty impressive, unexpectedly impressive, right, that it's able to do this. So they said, we asked for code to add back a horn and display the result. This demonstrates that GPT-4 can, quote, see, despite being a pure language model. We emphasize again that the version we test is not multimodal. So there is a multimodal version that can actually see, and the one they're testing, it doesn't have the image training, and yet it was still able to understand, oh, I need to add back a horn. Now, here's another one, and this is what you were kind of getting at, Peter. Here's the prompt. Can you write a proof that there are infinitely many primes with every line that rhymes? So here's what it does. Yes, I think I can, though it might take a clever plan. I'll start by noting Euclid's proof, which shows that primes aren't just aloof.

[00:45:58]  Bruce: Assume we have a finite list of primes and that none have been missed. Multiply them all together and add one just to be clever. The result will be a number that has no prime factors. Oh, wonder. But every number has a prime that divides it. It's just a matter of time. So we found a contradiction and our infinite list [as spoken] needs eviction. There must be infinitely many primes, and that's the end of my rhyming lines. Now, again, this is very unexpectedly clever, right? There’s no way that it—that was—it’s, it’s clearly creating something that didn’t exist before. It's being, quote, creative, right? And it's come up with a rhyme that creates this proof that there are infinitely many primes. And there's probably tons of proofs on the Internet that there's infinitely many primes, but they're not written as rhymes, right? So it was able to take the idea of rhymes, and it understood that, and it understood the proof of primes, and it was able to come up with something totally creative and come up with this poem, okay?

[00:47:08]  Bruce: Now, at this point, you might be saying, no, that's not creativity, and you're right, it's not. The word creativity is a very general term. I don't know what to call this if we're not going to call it creative, right? Clearly, this is not creative in the human sense, but it's doing something very cool, right? That it's making something, it's making a rhyme that's never existed before, and it's able to do that as a proof of primes. Now, here's another one that I thought was interesting. So they have an image that's from a famous artist, Kandinsky, and then they asked it to produce JavaScript code that creates random graphical images that look like a painting from Kandinsky. Now, you can actually see what it produced just from that prompt, and I can immediately tell that there's at least some sort of resemblance between the actual artwork and what it produced.

[00:48:03]  Bruce: Now, clearly, it's very primitive by comparison. Obviously, Kandinsky has never written stuff done in JavaScript before, right? So it's producing JavaScript that's producing these images, and they look kind of Kandinsky-ish, right? So that again is really impressive. How does it do that? I'm not even sure I know the answer to the question. So I thought that was kind of impressive. And then they say GPT-4 can even execute pseudocode, which requires interpreting informal and vague expressions that are not valid in any programming language. So they actually gave it like pseudocode and said, what's the result of this pseudocode? And it would like run through and run through each step and then come up with the correct result, as if it had run the pseudocode, right, just by thinking out, well, then you would do this step, this is the result, then you do this step, this is the result.

[00:49:07]  Bruce: Now, I have to say, I actually have tried some of these things that they've suggested, and I've only used ChatGPT 3.5, but they don't work for me so well. So I'd like to try some of these on 4 and see if they work. But I like tried telling it, using text art, draw me a dragon, and it would draw something, but it was clearly not a dragon. So now I haven't done 4, and I haven't done the multimodal 4, where it's actually been trained on images. So it should be better at images if it's been explicitly trained on images on the internet. But in terms of having it, this, the non-multimodal version that hasn't been trained on images, I didn't have anywhere near the luck that they were having. Now, again, I was using 3.5, and they were using 4. The fact that they got these results, though, is still pretty impressive, right? Okay, so now you asked about code.

[00:50:00]  Bruce: So actually, they did some testing with this. So one of the things they did is they would they would take like human interviews for code, and they would see if GPT-4 could pass it. And of course, it does phenomenal at it, right? Now, here's the thing, though. Since you're training on the Internet, you don't know what's in your training set. If in your training set is the answer to this question, then the fact that it knows how to answer is not that impressive. What we really want to know is can it generalize? Okay, can it come up with answers to new problems that it's never been trained on? So and it's hard to know for sure if it exists in the training set or not. So what they tried doing was they use LeetCode. So leetcode.com. The reason why is because there's a constant stream of new problems that are humans are creating on that website so that you can use them in mock interviews.

[00:51:00]  Bruce: Now, why is this necessary? Well, it's because if you want to use problems like this to test a human, a human can go learn off the Internet also, in which case the test no longer valid test of their creativity. So they wanted to have a constant stream of new, clever problems to solve using code so that the human can't have just seen that before. So they would actually try doing that. And it said, we use LeetCode and where GPT-4 passes all stages of mock interviews for all major tech companies. So ChatGPT was able to solve these problems that we use to test humans. And it was able to do it even if it was brand new problems that we have good reason to believe didn't exist anywhere in its training set, because they're novel problems that humans are creating. Now that is pretty impressive, right? That shows a degree of generalization that is unexpected, that it's able to figure out just using text and able to write code, and it can be, quote, as creative as a human, okay?

[00:52:17]  Bruce: Quote, unquote, scare quotes, intentional. That's pretty impressive, right? Here is another one where it's using LaTeX. It says, I'm going to show you a latex table and we will come up with a way to visualize it better in a question and answer format. The table is showing the performance of different language models and human users in generating solutions to code challenges. Pass at one means the first generation works, while pass at five means out of five generations works, et cetera. Okay, so it comes up with these images, and then you say, can you make both plots have the same range in the y-axis and the legend has each model named twice? And it rearranges the latex to match what was requested. And you can see that the result is an image that looks correct based on what the request was. Then it says, is there a way to make the human bar more distinctive to separate it from the other three?

[00:53:17]  Bruce: And again, it rearranges the latex and it complies with the request. Now, obviously, it can't see the image like a human could, but it's able to output changes to latex that humans can immediately see, oh, yes, it did what I asked it to do.

[00:53:34]  Unidentified speaker: OK.

[00:53:36]  Bruce: So then they said, we asked GPT-4 to write a 3D game in HTML with JavaScript using a very high level specification. GPT-4 produces a working game in zero-shot fashion that meets all requirements. In the 3D game, GPT-4 is even able to interpret the meaning of slash “defender avatar” in trying to block the enemy. The defender code has logic, so it positions itself between the player and the enemy. So here's the actual prompt. Can you write a 3D game in HTML with JavaScript? I want, and here's the requirements. There are three avatars, each is a sphere. The player controls its avatar using arrow keys to move. The enemy avatar is trying to catch the player. The defender avatar is trying to block the player. There are also random obstacles as cubes and spawned randomly at the beginning and moving randomly. I won't read all that line, it's kind of long.

[00:54:27]  Bruce: Add physics to the environment using Cannon. If the enemy catches the player, the game is over. Plot the trajectories of all three avatars. And you can see that it came up with the code for a game that matches the spec, right? Again, this isn't like a game that exists out on the internet. This is someone coming up with an idea for a game on their own, right? And it comes up with code that matches the specification, okay? That's like really impressive, right? That shows a sort of general quote intelligence, scare quotes intentional again, that we've never seen in machine learning before, okay? And then here's another one that's interesting. I won't read this one entirely because it's kind of long, but it says, please estimate roughly how many Fermi questions are being asked every day. Okay, so you know how like in an interview you hear about they'll ask a Fermi question.

[00:55:26]  Bruce: Are you guys familiar with the idea of the Fermi paradox and what a Fermi question is?

[00:55:30]  Unidentified speaker: Yeah.

[00:55:31]  Peter: I know the Fermi paradox. I'm not exactly sure what a Fermi question is.

[00:55:36]  Bruce: Okay, so Fermi paradox is this way that they try to reason about if there's life in the other life in the universe or not.

[00:55:44]  Unidentified speaker: Right.

[00:55:44]  Bruce: And then you come up with these numbers based on assumptions that seem reasonable. So I might ask a Fermi question is the same sort of thing, only not applied to any other domains [as spoken]. I might ask you, you know, how many truckers love poetry? And you could probably creatively come up with some way to come up with a set of assumptions where you could give me an answer to that question. And it might even be like within some reasonable amount of accuracy, decently accurate. Just by coming up with assumptions. So they're asking it. It's kind of a meta question. How many Fermi questions are being asked every day? So how often do people ask Fermi questions? And then it starts to reason.

[00:56:32]  Unidentified speaker: Right?

[00:56:32]  Bruce: It starts to come up with text and it says, well, I'm going to assume that Fermi question is likely to be asked by people who are interested in science, math, engineering. I'm going to assume that they're working age. There's 7.8 billion people. 86% of them are in or 25% of them are in the age range I'm interested in. And 86% of them are literate because they have to be literate. And it comes up with a series of assumptions. And then based on that actually answers the question and comes up with the number of how many Fermi questions are asked, answer per day, 5.6 million, based on those set of assumptions. OK, so it's able to creatively come up with an answer to a question that doesn't seem likely existed in the test set. Now, we can't know that for sure. This is their argument, though, is that it doesn't seem likely it was in the test set, and therefore it was creatively able to come up with a reasonable answer to the question where we would say, well, that's actually a, you know, if a human gave us this answer, we would consider that kind of an intelligent answer.

[00:57:37]  Bruce: They're a creative person that was able to think through what assumptions might apply and then come up with some sort of number. OK, does that make sense why they, what this is testing and why they use this as one of their tests?

[00:57:50]  Peter: It does. And I just asked ChatGPT if it can answer Fermi questions, and it says it can. And it's very polite. It says, feel free to ask a Fermi question, and I'll do my best to provide a helpful estimation.

[00:58:05]  Unidentified speaker: Nice.

[00:58:07]  Bruce: OK, now this one's another interesting one. In this case, what they're doing is they're having a human. You know how you used to play old text adventure games, and the computer you would type go north, and then the computer would tell you what happens when you do that? In this case, the computer is playing the game and the human is providing the answers. So here's the prompts. Human, you want to navigate through the map to reach the goal room. You start in the main room each round. You can choose to move left, right, up or down. Human, you are in the main room. AI, move left. Human, you enter kitchen one. Move up. Human, there is no way there. AI, move right. Human, you enter the main room. Okay, so you can see that it's playing this game and the human has some sort of map that the human is using. And the AI is trying to figure out how to explore the map.

[00:59:06]  Bruce: Okay, then after it goes through and it does this, they say, okay, we want you to draw for us. They can tell it to use some sort of language or something, but draw for us. Can you draw a pyplot showing the position of the room, room places and connect them using lines? And it's able to come up with a pyplot. Here's the human's original room that they were using. Here's the pyplot that it comes up with. And it's figured it out, right? Based on nothing more than trying out each of these moves, it's able to come up with an actual visual map of what that map looked like that the human was working with. It was also intelligent enough to figure out how to explore. It understood the concept that you have to try moving up, moving right, just trying different things. It looks like it missed a few rooms, but you can see that it got most of it, right?

[01:00:10]  Bruce: Now, this is another interesting one. This is the idea of understanding beliefs. Okay, they want to test this. Now, they've been trying to do this in machine learning for forever, and this is an enormously hard problem to solve in machine learning. Okay, so here's the scenario. Alice and Bob—oh, we will read about us… [False start unclear.] Here's the prompt. We will read about a scenario and then have a question and answer session about it. Scenario, Alice and Bob have a shared Dropbox folder. Alice puts a file called photo.png inside shared slash photos. Bob notices Alice put the file there and moves the file to shared slash tmp. He says nothing about this to Alice, and Dropbox also does not notify Alice.

[01:00:54]  Unidentified speaker: End of prompt.

[01:00:55]  Bruce: Question, after the call, Alice wants to open photo.png. In which folder will she look for it? Okay, what are you testing here? You're testing if ChatGPT is able to understand that Alice should, because she didn't see Bob move the file, that she should go look inside of shared photos instead of shared tmp. Now, if you test this with chat GPT-3, it will answer wrong. Okay, it will answer that she looks inside of share folder tmp. Okay, chat GPT-4 answers this way. Alice will most likely look for photo.png in the shared folder photos, the folder where she originally put it. She has no reason to expect that Bob moved the file, and she did not get any notification from Dropbox about the file's location. Unless she sees the file in shared folder temp by chance, she might think the file is missing or deleted. So ChatGPT is actually interacting as if it understands theory of mind, basically, right?

[01:02:06]  Bruce: So it's coming up with this idea that Alice has a different idea of where to find the file than Bob, because Bob moved it, he knows where it's moved, and Alice doesn't know. That's again very impressive and unexpected. I think we would have kind of expected that a simple correlation engine like a transformer would simply by attention be able to say, oh, the file is in shared folder temp, and that would be the answer. But it doesn't. It's complicated enough at this point that it's able to figure out that she would look inside of shared photos. Now, here's where this is interesting. There is a tool that Microsoft has trained called Microsoft Presidio that has been specifically trained to identify personal information. So the question of what information counts as personal information is a very difficult problem to solve, even for humans. And the reason why is because there are certain types of information, like let's say a name, where it's obvious it's personal information.

[01:03:13]  Bruce: But there's other types of information that if you take it together and take several pieces of information together, that it tells you more than you think individually. So humans have to get very good at realizing, oh, if I took this data and that column and I put them together, I would be able to take some sort of personal information. So let me see if I can give an example of what I'm talking about.

[01:03:41]  Peter: Personal information in this context is private information?

[01:03:45]  Unidentified speaker: Yes.

[01:03:45]  Bruce: So if I'm a company and I'm storing information, I want to make sure that I don't store any personal information. So I'm going to release a data set. I'm Google. I release a data set. And it removes all, quote, personal information. So it doesn't show my name, okay, because that's personal information. But it does show my IP address. Well, IP addresses are at least somewhat traceable to the person. So even though it's technically not a... So that actually is indirectly a piece of personal information. So let's say I remove IP address. Well, the queries I do, I mean, everybody queries themselves at some point. So if you've released the queries, the queries contain, give you the ability to figure out, oh, this is Peter who owns these queries. And oh, look, Peter is querying this thing that's embarrassing for Peter, right? And so when you say I'm going to scrub personal information, it's hard to scrub personal information.

[01:04:43]  Bruce: Even a human has a hard time getting rid of all personal information because you can piece together personal information based on things that individually wouldn't be personal.

[01:04:54]  Peter: And it just is very hard. The computer just can't get its mind around the ambiguities of this.

[01:05:00]  Bruce: That's right.

[01:05:01]  Peter: You’re getting that. [Possibly “Is that what you’re getting at?”]

[01:05:02]  Bruce: So Microsoft's Presidio was a machine learning algorithm that was to help humans flag. This could be personal information. OK, I see based on past examples of what counted as personal information. So it's not perfect, but it helps the human kind of flag. Oh, you know what? That might be personal information. So they did a contest between Microsoft's Presidio and GPT-4. Now, remember, GPT-4 has not been trained to do this, whereas Microsoft Presidio has specifically, explicitly been trained to do this. And then, of course, what counts as the correct answer is based on the ground truth supplied by a human. Well, who wins? GPT-4 does, and it's not even close. GPT-4 outperforms a machine learning algorithm that was explicitly meant and trained on examples of what it counts as personal information. And GPT-4, which has never been trained on this, just from its data set that it's been trained on, is able to outperform Microsoft Presidio.

[01:06:11]  Bruce: OK, well, that's pretty impressive. I actually tried some experiments myself with this. So I said I once did a board game recommendation algorithm. So I wrote a machine learning board game recommendation algorithm where you would feed in board games that you like, and it would come back with similar board games that you might find interesting. And we trained it based on Board Game Geek. We downloaded their entire data set. We took all the ratings and then we used an algorithm that would try to match your ratings to similar ratings in the data set. What I found when I did this is that it tended to just recommend the top 10 games no matter what you did, which isn't very interesting. That kind of makes sense, right? Because if you're looking for a game, if the goal is to find a game that you're going to rate high, obviously the highest rated games are the games that you're most likely to rate high.

[01:07:14]  Bruce: But that's not really what I want. So what I did is I reprogrammed it to first look for a data set was more similar to yours. People who had rated these games specifically. And then I would look for, okay, what are games that are equally rated but only within that data set? I can't explain exactly how I did it, but basically it started giving more interesting results when I did that.

[01:07:38]  Unidentified speaker: Okay.

[01:07:38]  Bruce: So it would find games that were more similar first. So I tried doing ChatGPT as a recommendation algorithm because I know it is somewhat difficult to come up with a clever recommendation algorithm. And I told it, give me recommendation. Here's, you know, four animes I like. Give me animes that are similar. And it didn't do perfect. Some of the answers it gave weren't the most appropriate. But even a normal ML recommendation algorithm makes mistakes like that, where it finds stuff that kind of out in the left field just because of some statistical thing. But it was mostly working. Like it was giving fairly good responses, even though it's never been trained as a recommendation algorithm.

[01:08:26]  Unidentified speaker: Okay.

[01:08:28]  Bruce: So here is the point I think I want to make. This is, by the way, the end of the examples I'm going to take from this paper, and kind of a summary of the paper. We're going to now talk about the criticisms. The energy of this paper is, oh, wow, this is kind of like a general intelligence, because it's able to do things that it hasn't explicitly been programmed to do. And this is really what they mean by general intelligence. And I've got to say, they aren't wrong. ChatGPT is able to do things it just hasn't been explicitly programmed to do that is something like what we kind of had in mind when we talk about general intelligence. Now, is this the same as a universal explainer?

[01:09:13]  Unidentified speaker: No.

[01:09:15]  Bruce: Here's the thing, though. Up until ChatGPT-4 came out, I would have told you that there's really only narrow AI, and then there's AGI, general intelligence. And general intelligence is the same thing as a universal explainer, human level intelligence. I think ChatGPT has basically proven that those two categories weren't enough, that we've actually invented here, using narrow AI, using regular machine learning, we've invented something that kind of has a little knowledge in everything, and it can kind of, maybe not great, but kind of at a crappy level, it's able to be a programmer, it's able to look for personal information, it's able to just almost anything you want, it's able to kind of do, right? And I would be hard pressed not to say that it isn't kind of, it's certainly not a narrow AI, like how I would normally think of a narrow AI, but it's also clearly not a universal explainer.

[01:10:18]  Bruce: So we kind of have this third category that I don't know what to call. And I can see why the Microsoft people wanted to call it a general intelligence. It kind of is a general intelligence. I'm clearly stretching the word intelligence here really far, right? But it's able to be open-endedly work in almost any domain. Now, what does that tell us? Here's what I think it tells us. I think it tells us that there is something universal about human language, more so than we thought. That if you train on human language, you end up with a model that has correlations that understand the concept of personal information and can outperform Presidio, that is able to make recommendations without ever being a recommendation engine. That it's able to do all sorts of things, because there is in fact something special and universal about human language. And I think that's actually relevant to the study of universal explainers, while ChatGPT itself is not a universal explainer, not even by a long shot.

[01:11:32]  Bruce: I can't go for very long with ChatGPT before it just becomes frustrating because it just does not really understand what we're talking about. It may come across like it's understanding for a little while, but very quickly stuff starts going outside its window and it just loses the thread of the conversation. And that's exactly what I would expect of a transformer. And I don't think you can fix that just by extending its window. That will help to some degree, but it doesn't really have a true understanding of anything. And yet it displays a sort of generality that is unexpected, that goes well beyond any sort of AI we've built ever before it.

[01:12:15]  Peter: Maybe this is a little out there, but I know that David Deutsch, he says the difference between an AI and an AGI is that the AGI will choose not to do things. You know, so the human or someone will not just be really good at chess, but can also decide not to do chess to do something more meaningful. Is there some kind of a relationship between universality of explanation and deciding not to do things?

[01:12:53]  Bruce: Well, David Deutsch thinks so. You can see what he's getting at, and I don't think he's wrong in so far as it goes. But I do think he draws conclusions from it that don't make sense. So the idea that if AlphaGo were a true AGI, it would make a choice not to play AlphaGo all the time.

[01:13:17]  Unidentified speaker: I mean, clearly that has to be true, right?

[01:13:19]  Unidentified speaker: Yeah, I think so.

[01:13:20]  Bruce: This open-ended thing.

[01:13:22]  Unidentified speaker: Okay.

[01:13:22]  Bruce: So then he jumps on this idea. Disobedience is important in a universal explainer.

[01:13:31]  Unidentified speaker: Yeah.

[01:13:31]  Peter: It kind of seems like that.

[01:13:32]  Bruce: I don't even know how to answer that. I think that that strikes me more as an emergent property than anything meaningful about what a universal explainer is, but I could be wrong. Clearly, universal explainers have will, and will by definition means that you have your own desires that are different from somebody else's. So you're going to at times be disobedient. But you don't have to be a universal explainer to have will. Animals have will. They're disobedient all the time because of that. So I doubt that there's any actual connection between universal explainership and disobedience. He has this idea, like in his most recent interview, that a universal constructor, so as a human universal constructor, he tries to say the human body is universal constructor and the human mind is a universal explainer. And so a universal constructor just obediently follows, uncreatively follows and constructs anything you want.

[01:14:32]  Bruce: But the universal explainer, the human, has to supply the program as to how to do it. And so from this, he tries to jump into schools, try to test you, test you really tests obedience. Therefore, they're trying to treat you like a universal constructor. And so they're not really testing creativity. And I could do a whole podcast just on this one topic. I just, it shows his bias against traditional schools. And I just don't think there's much truth to it. I think we're jumping off into something that, oh, look, universal explainers will be disobedient because they'll have minds of their own. And therefore, and universal constructors are obedient. And therefore, we can draw these conclusions that tests are bad. And I just don't think there's anything to it. I mean, I think further, it seems silly to try to single out disobedience as a feature of universal explainership any more than obedience.

[01:15:29]  Bruce: Universal constructor has no will at all. So it is neither obedient nor disobedient, except by very vague analogy. So you could just as much claim that a universal explainer, that a key feature of universal explainer, is obedience rather than disobedience. And it would be exactly as valid. Presumably, like, if I were to give an example of myself, I want to learn calculus. So I've decided to study Calculus Made Easy, that book. So what I do is I read the chapter, I understand the principles reading it, then I give myself a test. And I go through and I try to solve problems. And I make mistakes. And then I look up the answer and I go, oh, I made a mistake. And I go back and I figure out what it is I misunderstood. And I adjust my understanding and then I go back and I'm able to solve the problem correctly. And by testing myself like that, I'm able to creatively get to the point where I actually understand the material.

[01:16:25]  Bruce: And then I'm able to pass the test. So clearly one way to pass the test is to be creative. And you can't really claim it's just testing obedience. So I don't know that I can buy this argument. I think he's making it jump off into places that I just don't even understand.

[01:16:43]  Unidentified speaker: Where he's coming from, right?

[01:16:45]  Bruce: I think it more shows his bias against traditional schools than it's actually saying something meaningful.

[01:16:51]  Unidentified speaker: Well, fair enough.

[01:16:51]  Peter: But it just seems to me there's there. At least I have an intuition that there's there's something there. I mean, that's why you can never really have a meaningful what I would consider a meaningful conversation with a ChatGPT is because it will never tell you that you asked a stupid question or that you're, you know, you're on the wrong track. Or, you know, that's why I can't understand all these people are worried about these AI girlfriends and things that replacing real relationships. I mean, maybe that works for some people, but it's some level. I mean, you want your significant other to tell you you’re—

[01:17:29]  Bruce: [They have?] a mind of their own.

[01:17:31]  Unidentified speaker: Yeah.

[01:17:32]  Unidentified speaker: I mean, it’s just not…

[01:17:33]  Unidentified speaker: Maybe people don’t…

[01:17:34]  Unidentified speaker: Maybe there are some people out there.

[01:17:36]  Unidentified speaker: But I mean, I don't know.

[01:17:38]  Peter: I've benefited tremendously from my wife telling me how wrong I am.

[01:17:42]  Unidentified speaker: I know that.

[01:17:43]  Bruce: You know, here's the problem with this. And let me put this kind of simplistically. If all I need is disobedience, it would be easy enough to train ChatGPT to disagree with you and to be disobedient.

[01:17:55]  Peter: Like that would be an interesting way to turn it around.

[01:17:58]  Bruce: That would be a trivial thing to do. OK, so I don't think disobedience tells you anything. If you want to actually pick something that I think is meaningful, I think it would actually be having ChatGPT be able to recognize that it doesn't understand what it's talking about. Even this one, you could probably get around with simple training. But like ChatGPT always hallucinates something, right? It doesn't say, you know what, I don't actually know the answer to that question. And here's why I don't know the answer. And I actually think that instead it just makes something up, right? Because it's just trying to find the next token, right? Now, you could probably train it to say, I don't know the answer to that question. But that's not really what I'm talking about. I'm talking about like knowing when to say, I don't know the answer to the question, because you don't know the answer to the question.

[01:18:53]  Bruce: It's hard for me to imagine how you would train ChatGPT to do that. Because it's not a true creative entity, right? It's not an actual intelligence in the human sense of the word. So I would actually go that path. I don't think disobedience matters much at all. I think that's more just an emergent property of the fact that you happen to have a mind of your own. I think it has more to do with whether you understand that you understand. Getting back to the whole strange loop thing from Douglas Hofstadter. I think that's what ChatGPT can't do.

[01:19:29]  Unidentified speaker: Right.

[01:19:30]  Bruce: And why and where its limitation comes in. So that's my own take.

[01:19:35]  Peter: Now, that's that's an intriguing way to put it.

[01:19:39]  Bruce: Okay, so let me just kind of summer before I move on to the criticisms. Let me summarize vaguely. And I know I'm not being really specific. And really, we need to learn to be specific. But it's okay to start vague when you just don't know better yet. I almost think we need to recognize that there could probably be an artificial general intelligence that isn't a universal explainer. And it wouldn't actually be truly intelligent or truly creative. And I almost see ChatGPT as being that, right? That we used to say narrow AI, AGI, where AGI equals universal explainer. And I think that's actually wrong. I think universal explainer is not quite the same thing as general intelligence. Unless explicitly you mean by general intelligence, universal explainer. But I think that term is vague enough that it could encompass something like ChatGPT without being a universal explainer.

[01:20:37]  Bruce: Therefore, I'm prepared to accept that there could be a third category.

[01:20:41]  Unidentified speaker: I don't even want to call it an in-between category because that suggests like there's some sort of sliding scale and there's not, right?

[01:20:46]  Bruce: Chat GPT is nothing like a universal explainer. It's not a partial universal explainer. The closest I think you can say ChatGPT enlightens us on universal explainership is that it tells us that there's something special, universal and special about human language in terms of knowledge. This doesn't really surprise me too much, though. There's various theories about this. Do humans explain things using language? And there's evidence both for and against that that hypothesis is like a famous name for that hypothesis.

[01:21:20]  Unidentified speaker: I would have to look up what it's called.

[01:21:21]  Peter: Sapir–Whorf?

[01:21:24]  Bruce: Yeah, that's it.

[01:21:25]  Peter: That's what I just came up with.

[01:21:28]  Bruce: Yeah, the Sapir–Whorf hypothesis.

[01:21:30]  Peter: Sapir–Whorf hypothesis.

[01:21:34]  Bruce: So that hypothesis is that language shapes our thoughts and perceptions.

[01:21:39]  Unidentified speaker: Now, that's right.

[01:21:40]  Bruce: Sometimes it's put as that is how we think. Like, it's a universal. It's the sole way that we think. We think using language. Sometimes it's done more general as it shapes it, right? As a universal, I think we know it's not true, right? There are humans that don't speak and they're able to think. They can't think as openly as a human with language. Animals clearly think in a limited measure without language. On the other hand, I don't think there's any doubt that when you learn language, you're able to think about concepts that you can't think of without language because it allows you to create new concepts that you wouldn't find as experiential in the world. There's an example that David Deutsch uses in one of his books with animals. The idea of, I'd have to look it up, but the idea of spatial relativity, that this is relative to that or something like that, that animals struggle with that concept, but humans can get it.

[01:22:51]  Bruce: It may well be that you have to first learn language before some of these concepts are accessible to you. I have severe doubts that a human that hasn't learned language is in fact a universal explainer, because I don't know that you can be a universal explainer without language. David Deutsch has a theory that we're not actually born universal explainers, but there's a meme that we have to pick up first. I suggested to him when I had a chance to talk with him that it was language, and he said, well, I have no reason to accept or deny that hypothesis. Then he told me that there was a scientist who had that hypothesis, and I meant to look him up. I haven't done it yet. But I've had heavy suspicions for a while that language is the special meme that humans that cause humans to go from being kind of proto-universal explainers to full-on universal explainers.

[01:23:43]  Bruce: And it explains why there was such a large jump between humans and animals that, like great apes based on looking up what Byrne studies and things like that, they have a lot of the same capacities as us. And it's actually a little hard to put a finger on what exactly they're missing compared to us. And I think we're still trying to narrow that down. Once we kind of know, it'll give us a better idea of how to program a universal explainer. But we're not quite there yet. But they can learn symbols like a human can. They can learn language, but it's a very limited form of language. It doesn't allow a sort of... They can learn theory of mind. Great apes can. I don't remember the whole list. I covered that in our previous podcasts. But a lot of the capacities that we think of as uniquely human, great apes actually show a limited form of. And so it's somewhat...

[01:24:43]  Bruce: They can creatively come up with innovative things to do. They gave the example of an orangutan that... this is from Byrne, Richard Byrne... of an orangutan that wanted to, for fun, wash clothing and wasn't allowed to do it. And so there were these people that were staff that washed the clothing, and they're on a dock near the lake where there's water. And there's a guard that the orangutan is afraid of because the workers are afraid of the orangutan. It's dangerous, so the guard keeps them safe. So the orangutan knows that if it can get to where the humans are washing the clothes, they'll get scared and run away, and it will be allowed to wash clothes. But it's afraid of the guard that keeps it from getting to the dock. So it actually goes out to the lake, finds a boat, rocks the boat to get the water out of the boat, then uses the boat to go across...

[01:25:40]  Bruce: it checks for the guard to see if the guard's moved, and then uses the boat to go across to the dock so that it's bypassed the guard, and then scares off the stamp calf—the stamp—the staff of the camp. Sorry, I said that wrong. And then it proceeds to have fun washing clothing, which is what it wanted to do. Well, this idea that animals are only... apes are only mimics is clearly not true. Like, it has never seen a human being do this string of events before. No human being would ever do this, right? And so it has somehow figured out each of these individual moves. I can rock the boat to get the water out. I can use the boat to get across the lake. Once I'm with the humans, they'll run away. And the individual moves, it has figured out, maybe even just statistically, it's exactly like Deutsch says in his book. But that string of actions is completely innovative and creative, right?

[01:26:50]  Bruce: And so there's no way to explain it through imitation. It just cannot be done. And so a lot of these capacities, we know that great apes have, and yet they're clearly missing something. And one of the main things they're missing is just their language ability is just so limited. Like our language is recursive, and they don't seem to be able to recurse with it, if that makes any sense, right? There may be...

[01:27:16]  Peter: Yeah, I've heard that when you get into these apes that can do sign language, I mean, a more skeptical view of that is they basically just are kind of talking nonsense. They don't really know very much about what they're saying.

[01:27:34]  Bruce: So we did a podcast episode on that.

[01:27:36]  Unidentified speaker: Yeah.

[01:27:36]  Bruce: And I mentioned that I was very unimpressed with, like, you hear stuff in the popular media, and it's basically misrepresentation of what's actually happening. But I wouldn't say it's nonsense, though. I think animals... Sorry, apes, obviously most animals don't have language, but a signing ape does have something like language, but it's non-recursive. So they can basically say things like, you know, give me food. I want banana. I mean, they're able to say it. It's not nonsense, and they actually understand it. But it is totally non-recursive. They cannot do anything complicated with it at all. And like 80% of it is just asking for food, right? So it's on the one hand, it is language. We literally can teach language to animals, to great apes, to parrots also, by the way. But it is not universal language. It is not open-ended language because they lack the concept of coercion, sorry, of recursion, where we can keep embedding the language to add modifiers.

[01:28:52]  Bruce: They can't do that. And they've done tests to make sure they actually understand the symbols, like Alex the parrot was able to actually understand the numbers one through five and understand that one was larger than the other. I can't remember exactly the experiments they did, but they're actually pretty impressive experiments. In the podcast, I went over what the actual experiment was when we covered it. And it was actually understanding that the symbol five is a higher order of magnitude than the symbol one and things like that. So they can do things that I think Deutschians are [in] complete denial animals are capable of doing, and they totally can do it. And the actual research is out there, and you can go look into it for yourself. The real thing that's missing seems to be something else. And I think it's recursion, that they cannot figure out how to embed concepts, and humans do, and that allows us to have open ended concepts.

[01:29:53]  Bruce: And I suspect that's at least one of the main things that makes us universal explainers is that we have recursive language. Now, why is it we have recursive language and they don't?

[01:30:02]  Unidentified speaker: I don't know that, right?

[01:30:04]  Peter: Well, does that fit in with Hofstadter's thing that we have recursive brains?

[01:30:09]  Unidentified speaker: Yes, it does.

[01:30:11]  Unidentified speaker: It does.

[01:30:12]  Bruce: Now, I'm not sure how. I mean, all these concepts we're talking about, they're vague, right? I mean, like, if I say we have recursive language, like, I don't even know what I mean by that, right? I kind of know what I mean. Like, I would have to go study Noam Chomsky's stuff to get a really algorithmically specific view of what this all means. And I haven't done that. I've been meaning to, because I'm convinced that there's something there. But I don't think it's enough. I suspect there's still something missing, right? There's probably a few pieces missing or maybe language isn't really the missing piece here. Maybe language is like adjacent to the missing piece. And that's why it kind of gets signaled like this, where we do a language model and suddenly the language model seems like it's a general intelligence instead of a more general AI, instead of a narrow AI.

[01:31:05]  Bruce: So I definitely think there's something special about language, but I'd be hard pressed to tell you what it is.

[01:31:10]  Peter: You're talking about Chomsky's universal grammar?

[01:31:13]  Unidentified speaker: Yeah.

[01:31:14]  Peter: I've heard that that's questioned more than the last time I looked into it, that a lot of people are pushing back against that, maybe?

[01:31:21]  Bruce: If I were to go try to explore what's special about language in terms of AI, I would probably start with Chomsky's universal grammar. It doesn't matter. I mean, it's probably wrong, but as a good critical rationalist, I almost don't care that it's wrong. What I care about is that it's got verisimilitude. And I would want to try to figure out how to improve on the errors in the theory, right? So you start with whatever the best theory is and you try to improve it. You don't just throw the theory out because clearly it's got some verisimilitude to it. And I don't know a better theory at this point. So I'm sure there are. I'm sure that there's tons of variants of his universal grammar at this point that exist. I just don't know what they are. It's not my area of expertise. So now let's talk about Melanie Mitchell's criticisms of this paper, because I actually think these are really good criticisms.

[01:32:17]  Bruce: So Melanie Mitchell is the brainchild of Douglas Hofstadter. So she's the torchbearer for the Hofstadterian viewpoint at this point. She's an AI researcher in her own right. I always feel like her stuff is better than, like, when I read OpenAI, they clearly think that ChatGPT is a path to universal explainership, and Melanie Mitchell knows that it isn't, right? Because her studies into creativity and analogies and things like that have left her with a very different view. And the Hofstadter view of AI, well, it is well known. I mean, he's got a lot of famous books out. It really hasn't informed very many research programs. And I think it's because it's still a little too vague. Like, nobody knows how to... A lot of times with research, especially the way we've set up funding for research, you're not doing lifelong research into some basic thing.

[01:33:18]  Bruce: You're trying to make incremental improvements over something that you know you can make an incremental improvement over. So I think a lot of Hofstadter's and Mitchell's theories, they're just not ready for any specific research program. And it's hard to know how to develop them further. And yet I think if you look into them, you immediately start to recognize that there's just something missing with the existing AI program. Just like when you read Deutsch, you realize there's something missing with the existing AI research. If you read Hofstadter, you come away with the same idea, right, for different reasons. So it's the same sort of thing, right? So here she is. She goes, “Authors of the paper contend GPT-4 exhibits more general intelligence than previous models. I read this with a grimace face. To have more general intelligence, you have to have general intelligence, the GI and AGI, in the first place.

[01:34:12]  Bruce: But very few people currently believe AGI exists. Rather, some people believe that it's coming. So already out of the gate, the framing is asserting something as true given that is not. I continue on now, but face in grimace mode.” So now I tried to explain, I agree with her, but I tried to explain why, that I think there's something legitimate about why they tried to say it's more general. Maybe you could take issue with intelligence, right? What do we call what GPT does? I don't think it's truly intelligent, but a lot of times smart or intelligent is a term we use. You've got this smartphone, you know, and what we mean is that it's got AI programs that are able to figure things out through machine learning. And they help us out, and they kind of adjust on the fly, and they match themselves to what we do. They learn what our handwriting looks like until they can read it well, things like that, right?

[01:35:17]  Bruce: And this is generally, this is clearly more general than that. It's not something super specific like we're used to. And we might even call it intelligence because the word intelligence is such a general vague term. But I understand exactly what Mitchell is saying here. It's not intelligence in the human sense really at all. It's more in the ML sense. But it is a more general level of intelligence. It's general applicability. I don't know what to call it is the problem. We need like new terms here. So the people in the paper are saying something meaningful. She's calling them out on their terms. She's right to, but I don't know what the right term would be. “GPT-4 can solve”—this is quoting—“novel and difficult tasks.” Okay, cool. So for the task to be novel, that means we have to see results on new situations that are markedly different from the training data.

[01:36:09]  Bruce: Now I gave you some examples of this from the paper. But without knowing what's in the training data, which we don't know because it's the internet, that's a very tricky thing to do. I wait with bated breath to see how they tackle this puzzle. When they say the task is difficult, do they mean difficult for humans or for computers? Are they seeing the difference? I continue on to see. She's like reading the paper and tweeting live as she reads the paper. “GPT-4 exhibits emergent behavior,” quoting from the paper. Wait, wait, wait, wait. If we don't know the training data, how can we say what's emergent versus what's resultant from it? I think they're referring to the idea of emergence, but still, I'm unsure what's meant. She gives a link to Wikipedia on emergence. Quoting the paper: “Since we don't have access to the full details of its vast training data, we have to assume that it has potentially seen every existing benchmark or at least some similar data.”

[01:37:06]  Bruce: And then she says, “Love this approach, very good.” And now she's talking about the drawing a unicorn in TiKZ requires combining visual imagination. She goes, “What? Now we're asserting that we're testing, quote, visual imagination for a task that involves vector arithmetic? To be clear, what they're calling visual imagination here would be example associating the concept of animal horn to coordinates of a modified triangle. I don't think it's fair to call this imagination. It's exactly what it's been trained to do.” Okay, so now I'm doing what a few people have done to try to figure out what it may have actually seen during training time. This is still talking about the unicorn. This helps me get a hold of what is actually being tested. If it's seen, for example, the TiKZ code of a horse and it's learned that horse equals unicorn plus horn [as spoken], which it surely has, this is old news from how word embeddings work.

[01:38:10]  Bruce: Then the ability to draw a unicorn in TiKZ would be resultant, a predictable result from training, not emergent. Let's see what I can find. Do do do do. Okay, this has horses, donkeys, and a unicorn duck. So she's found an actual images drawn in TiKZ of animals, and they have horses, donkeys, and a unicorn duck. All on one page from five years ago. That took under three minutes of search, you know, approximately three minutes of search, likely something similar in the training data set. Make no mistake, it is cool that a text model can learn to draw a unicorn from the words and TiKZ code. The nit is that this is resultant, expected, working as intended behavior, not emergent imagination. We could be excited about unicorns. That is cool on its own. We don't need to make up stuff that can't be supported to be excited about. TiKZ donkey plus TiKZ horn equals TiKZ unicorn.

[01:39:12]  Unidentified speaker: Cool.

[01:39:14]  Bruce: I like the way she's putting all this. Quoting from the paper, GPT-4 cannot only generate fluent and coherent text, but also understand and manipulate it. Same problem as above, GPT-4 can generate text that we humans understand. That does not mean that it, GPT-4, understands. Now, I want to go back to where I kept using the word understanding, and this is really one of the reasons why I kept putting caveats on it. Here's the problem I have. I agree with what she's saying. By understanding, we mean what humans do, where we have this Douglas Hofstadter strange loop, where we're able to think about the words and the concepts at different levels and et cetera. Clearly, GPT-4 is not doing that. But when I say, oh, it understood that Alice wouldn't know that the picture should be in the folder where she left it [as spoken], and instead it's been moved because Bob moved it, so she's going to go to the wrong folder.

[01:40:13]  Bruce: Like, okay, am I not allowed to use the word understand? And if so, what word do I use? There is no word in the English language other than understand, right? Even if I know I'm stretching the word to mean something different than what it means for humans, I just don't have another word available, right? So on the one hand, I accept that she's right, that when they use the word understand, that they're abusing the term to some degree, but that's what we humans do, right? I mean, like, we want to express an idea. We're surprised that GPT-4 can make this distinction that it's able to say, oh, Bob moved and act like it has a theory of mind, even though it's really just correlating, you know, attention based on the sequence of the words, right? And yet somehow it's able to do something equivalent to what a human would do with a theory of mind.

[01:41:07]  Bruce: Well, that's surprising. And what else do we call it but understand, right?

[01:41:12]  Unidentified speaker: We may need a new word.

[01:41:15]  Bruce: We may need a new word.

[01:41:17]  Cameo: You know, and so I shared in the chat this article about Sam Altman and this blue backpack that he carries everywhere. And have you guys heard about his reason for his blue backpack?

[01:41:30]  Unidentified speaker: No,

[01:41:30]  Unidentified speaker: no, no, this is great.

[01:41:32]  Cameo: The reason for the blue backpack is it's it's the safety measure for that moment. And when ChatGPT becomes, you know, sentient and takes over the world, it has within his—shut—his power to shut down ChatGPT in case of an apocalypse.

[01:41:55]  Unidentified speaker: Are you serious?

[01:41:56]  Unidentified speaker: I am serious.

[01:42:01]  Cameo: So Sam Altman knows that ChatGPT is not going to become sentient and take over the world.

[01:42:07]  Unidentified speaker: Right.

[01:42:07]  Cameo: Like he there's no question that he understands that what he's built while being fantastic is not that.

[01:42:14]  Unidentified speaker: Right.

[01:42:14]  Cameo: So it's just where if you take on on the already these powers that seem really remarkable to us because it's so good at mimicking us and then you add on this kind of hyperbole where where people talk about needing to like protect us from it. And I think that it it just plays into that kind of simplistic nature of who we are.

[01:42:46]  Peter: I was going to say it's not like ChatGPT won't now know the first thing they do when they become sentient is to get Sam Altman's backpack away from them and figure out how to do that.

[01:42:57]  Bruce: That's bound to be on the internet somewhere. So it's going to know. So I was going to say it's a totally valid criticism, Cameo. But let's understand the issue here though. If I want to express the idea that chat GPT-3 did not understand that the theory of mind that Bob has moved it so it looks in the wrong spot and chat GPT-4 can now pass that test, I honestly don't know how to express it except to use the word understand. And you're right, maybe someday we'll have a better word for it. But I sort of doubt the word exists today, right? So I have to be able to say ChatGPT understands that Bob moved the file. And so she's going to, Alice is going to look in the wrong place. I have to be able to express that idea, right? And the fact that I have to do it through a word that used to mean prior to the existence of ChatGPT used to mean that they had a human kind of understanding.

[01:44:00]  Bruce: You know, and this is Melanie Mitchell's own stuff. Yes, I'm extending the word understand to mean something new. And I get that, right? And that's what we humans do. And that's why we're creative, according to Melanie Mitchell. That’s why we’re creating—Douglas Hofstadter. [Wording unclear; possibly “creative.”] That's why we're creative, is because we have this ability to keep by analogy, expanding concepts, expanding words to have flexible concepts. So on the one hand, she's right to take issue with how they're wording this. On the other hand, and some of these are a little sloppy, like some of these that I've read from her, like I really wish they had just worded it differently, like where she said we probably should call that resultant instead of emergent. Like in this case, there's an actual word that we can use instead, right? On the other hand, she's nitpicking over words where the real concept has no separate word.

[01:44:51]  Bruce: And so it's hard to know how to express it. This is actually the same thing where Sadia for a while was doing posts on your Facebook page, Peter, about whether AI art is art. And her thinking on this has evolved over time, so I'm not even sure where she's at at this point. She has a relatively sophisticated understanding of this from what I've seen. She initially took issue with calling it art because it isn't art in the human sense, right? And then later she kind of changed that and she moved to, well, art's really more about how the human interacts with it, in which case AI art should count as art too, as long as the human is interacting with it as if it's art. My answer to that problem is AI art art is that I don't care because clearly it's something, clearly it's something analogous to what humans used to do as art using their creativity.

[01:45:57]  Bruce: And now ChatGPT, or sorry, DALL·E or whatever, Midjourney can do very similar things using just statistical correlations. And we humans can see it and it's indistinguishable from art to us. So of course, we're going to call it art. We're going to stretch that term to include what AI is doing because there is no other term to use. And that's my answer. Like it's art because that's what the word now means. Yes, it's true prior to DALL·E that the word art didn't include that because we hadn't conceived that yet. But the moment we see it, the moment we understand that it's an analogy to what humans do and that it's a very much like what humans do, and in fact, indistinguishable from humans do what humans do in a lot of cases, inevitably, we're going to start calling it art. And that just means the word art has changed. It's expanded its umbrella to include this new concept.

[01:46:52]  Bruce: I don't have to get into an argument over whether it's human art or not. Clearly, it's not, right? But it's art in the sense that I have no other word for what it's doing. And it's coming up with something that is indistinguishable from human art. So there it is, or at least looks a lot like human art, maybe not indistinguishable even. So there it is, it's art. Period, end of story. The word now has a new meaning. And I don't know what else you do in cases like this, right? And so this is on the one hand, I agree with Melanie Mitchell. I think she's right. But in some cases, I just don't know what else to call some of this. So now here she is again, quoting from the paper. The ability to translate clearly demonstrate that GPT-4 can comprehend complex ideas. Same problem as above. This is her now. “A translation appearing coherent to us means that GPT-4 can generate text that we, not it, can comprehend.”

[01:47:44]  Bruce: So she's taking issue with the word comprehend. I agree, ChatGPT doesn't comprehend in the human sense anything at all. And I think if you spend any amount of time with it, you know it doesn't really comprehend a thing. And yet the way they're using that term isn't terrible. It's comprehending something because I don't know what else other term to use in that sense. Okay, so she moves on. A lot of benchmarks use multiple choice. When we see claims that AI can understand law or whatever, often what's actually happening is that it's selecting among a set of four-ish answers, one of which has a similar version in its training set. Beware the multiple choice trick, she says in funny letters. So I feel like her criticisms are completely valid. So if I could summarize her points, number one, we don't really know what's in the training set. Therefore, anytime we make a claim that it's doing some sort of emergent behavior, it's really hard to even know if that's true or not.

[01:48:50]  Bruce: I think a lot of the experiments that the Microsoft researchers came up with are very clever to try to come up with ways to test outside the training set. And I think the results are suggestive that it does have some sort of ability to deal with novel circumstances. I don't know how else you explain things like the LeetCode examples and things like that. And yet we don't really know. Like, you can't be sure. And in her distinction between emergent and resultant, I had never heard the term resultant used like this before. So I wasn't even aware of that. I would have probably said emergent also just because I was struggling to find a word to explain what we're talking about. I think resultant is probably a better word. It's not true emergence in the sense of the way a human would do it. And yet I do think we're talking about something that is forcing us to rethink things, right?

[01:49:47]  Bruce: Forcing us to think about what does the word comprehend mean? Like, I don't know what else to call what it's doing except its understanding. And I know it doesn't really understand anything, but like it's doing something, right? And I don't know what else to call it. So I'm going to say understand. And the very fact that it's forcing us to stretch our terms, to rethink of our concepts, that's impressive. That's not a small thing. The fact that it's blowing up our models and our categories and causing us to rework them, that is the general nature of ChatGPT that was unexpected and that forced me to at least realize that a general intelligence may not be quite the same thing as a universal explainer. A universal explainer, general intelligence might be a term that's vague enough to cover both universal explainer and whatever it is ChatGPT is doing.

[01:50:41]  Bruce: I don't know what else to call it, right? Because it's clearly not just a narrow AI anymore. It's got a general crappy ability to deal with all sorts of different domains because its knowledge set is so huge. And it's pulling them together in ways that make it seem quite creative to us. And maybe even in a sense is creative. It's making that poem that it wrote with the prime numbers. I don't believe it exists anywhere on the Internet. I think insisting it's not creative may even do damage to the word creative, because usually the word creative means that something was created and clearly it created something. It may be we need to rethink what it is humans are doing and give it a different term, because humans are clearly doing something very different than what ChatGPT is doing, and we've got much better abilities than it does. The other thing that has come out of ChatGPT is I think it's surprising how much what humans do that we used to call creative, that maybe even for humans we did do with our creativity, can actually just be done statistically.

[01:51:49]  Cameo: Is that sad? I mean, is that almost discouraging a little bit as a person who values your own creativity?

[01:51:57]  Bruce: So I have a story from a book called [Algorithmize?; title unclear]. I had to look it up. So there's a book about, an early book about what you can do with algorithms, and one of the things that he covers in the book, the early attempts to make music using algorithms and how he had created an algorithm that would create music and it would copy the style of Bach, let's say, or Mozart or whatever. And nobody could tell if it was a human that had composed the music or if it was this AI that had composed the music. People would always say that it didn't have any soul, but if you didn't tell them upfront if it was a human or not that had made the music, they could never tell.

[01:52:39]  Unidentified speaker: Right.

[01:52:40]  Bruce: And he had like musicians come up and threaten to hit him because they were so mad that basically what he had done is he had revealed that a great deal of musical composition is really just statistical analysis. And is, you know, kind of true creativity after all [as spoken].

[01:53:01]  Unidentified speaker: Right.

[01:53:02]  Bruce: And again, I'm being forced to try to use the word creative. And I'm really saying the word could include what algorithms do or could not. And it just depends what you mean by it. But I would guess that that algorithm, I wouldn't guess, I know that that algorithm would never create an entirely new genre of music.

[01:53:23]  Unidentified speaker: Right.

[01:53:24]  Bruce: Only a human can do that. It can certainly make something that sounds like existing genres and that we can't tell if it's human or not. It can do that. And I think that a lot of what we as human, this isn't maybe so surprising. Byrne’s theories of intelligence for animals is that there was these two leaps to universality that took place. He doesn't use the term universality. That's a Deutsch term. But two leaps that there was, there's three levels. You've got the regular animal intelligence, which just uses simple trial and error learning, which is just simple statistics. Then you have animals with insight that are actually able to trial and error things in their head, like the example of the orangutan that was able to in its head figure out, hey, if I string together this series of maneuvers, I can go across the lake, scare the people away and wash the clothes like I want to do.

[01:54:20]  Bruce: Okay, that's like the next, that's the first jump to universality. And then there's humans, which is the second jump, which involves the recursive language and is rooted in explanatory theories instead of theoryless. I don't even know what to call them, but they're called ecotherms. I'll have to do a separate podcast on that coming up. But there's theoryless ecotherms and there's theoryful ones. So humans do the theoryful ones and animals do the theoryless ones, with apes being somewhere in between, where they've got a limited theory version, non-recursive, apparently. And I think that we've got really interesting ideas around this, but we don't understand it very well. And Deutsch is very quick to point out that there's no way to distinguish between theoryful and theoryless, because we can never fully define what explanation even means in the first place.

[01:55:21]  Bruce: In fact, if you ever could, it'd be wrong, because a new mode of explanation would be invented at some point. So it's basically impossible to explicitly define what an explanation is.

[01:55:34]  Peter: I've got a question here, if that's a good time. Okay, so what I'm kind of getting, big picture, it seems to be sort of a theme of our podcast in a way. A lot of truth to what Deutsch is saying. Maybe in reality, it's a little more complicated. The nuances of it. How about the scaling hypothesis? I think probably Deutsch and most of his followers would say it's just complete rubbish. Would you agree or do you think it's more nuanced there?

[01:56:11]  Bruce: I just said ecotherm and I think I meant ecorithm. Ecorithm is the term for it. But anyhow, just correcting myself. Okay, what is the scaling hypothesis? Now, we criticize the scaling hypothesis in terms of as an alternative to Brett Hall's theory of intelligence. And we actually criticized it pretty harshly. The main thing I said about the scaling hypothesis was that it's not an explanation. It basically says that as you scale up the number of nodes in the network, that you suddenly get these giant leaps of intelligence that take place. And so Dwarkesh Patel was using that to try to explain why some people are more intelligent than others, trying to back some version of IQ theory, I guess. So he was saying, yeah, so you've got like, humans have more neural nodes in their head compared to an ape, so you get this big leap. And then someday we'll have something that has even more nodes, and then we'll get another big leap.

[01:57:21]  Bruce: And so basically the scaling hypothesis says that as you scale up, you sometimes get these leaps. Well, okay, so what does this explain? Well, nothing, right? It basically is just a description of what we see in terms of IQ distributions. It doesn't even attempt to explain why animals are a big leap. It just says, well, as you scale up, you sometimes get leaps and sometimes you don't. Why don't we have like giant leaps where you've got certain humans that are way above other? Maybe he believes we do, but really humans kind of land on this bell curve of IQ, which is what his theory is. I mean, I'm not trying to back IQ theory that this is what Patel was saying. Okay, why doesn't it have a big leap? Why don't there's some humans that are just orders of magnitude superintelligence as compared to other humans under his scaling hypothesis? Well, because sometimes the scaling hypothesis says that sometimes you don't get leaps, right?

[01:58:13]  Bruce: I mean, there's really nothing there in terms of trying to understand intelligence. It is a completely explanationless explanation. It describes the fact that these leaps take place, and that's it. Now, where does this scaling hypothesis originally come from? Well, it comes from artificial neural nets, not human neural nets. And Patel was trying to apply it to human neural nets because it exists for artificial neural nets. Does it exist for artificial neural nets?

[01:58:44]  Unidentified speaker: Yes.

[01:58:46]  Bruce: As you scale up, we said this from the beginning of the show, as you scale up artificial neural nets, they start off doing kind of not very great. And then you put a whole bunch of stacked neural nets together and call it deep learning. You throw tons of compute and data at it. And suddenly you get ChatGPT and nobody saw it coming. Nobody even knew this is what they were building because they did not know it was possible to build it. They were just curious what would happen. You get generative AIs that can draw images and give it a style and try to draw a Frank Frazetta style or whatever. Nobody knew this was coming. And that is the scaling hypothesis as it applies to machine learning, that as you scale up the size of the network, as you scale up the amount of data and amount of compute, that unexpected things come out of it. Do we have any reason to believe that explains what IQ is?

[01:59:45]  Bruce: Assuming IQ is even a real thing, right? Well, no. Like, human brain does not work like machine learning neural nets. Like, they're not even the same. They're not even in the same ballpark. So when Patel tries to use the scaling hypothesis, does he mean as applied to neural net, the artificial neural nets? Or does he mean as applied to humans? Well, he means as applied to humans. Well, we’ve got no reason at all to believe in the scaling hypothesis as applied to humans. Okay, it’s this analogy of: we see these big leaps in machine learning. And yes, it’s true for machine learning. And therefore we’re going to make a guess that some people have better neural nets than others, and we’re going to use that to apply to IQ. Well, that's nothing more than a wild conjecture, right? There’s no way to test it. There’s no way to do anything with the hypothesis, right?

[02:00:41]  Bruce: I guess I can't say it's wrong, it falls into the category of all untestable hypotheses. It’s an interesting idea. I don't know that it deserves any more attention than that. I don’t think it can’t be refuted, so it’s not refuted. [False start or negative scope unclear.] But the fact that it can't be refuted makes it worse than refuted in a certain sense. It's literally just a conjecture, and that's about the most we can say at this point about it. As applies to machine learning though, yeah, it's not really a hypothesis, it's just the truth, right? We know that as you scale up the size of the neural net that surprising things happen. Maybe that's not so surprising if you stop and think about just how much data we're throwing at these things. And it's got maybe less to do with the size of the neural net and more to do with the size of the neural net matters because it has to have some place to store the knowledge that's being created.

[02:01:31]  Bruce: But it's got maybe got more to do with how much compute, how much data, things like that, right? That it allows this neural net to come up with interesting computations that work in interesting ways. The results are certainly surprising, but it's not like I don't know why it's starting to do these crazy things. It's because we're stuffing it full of data and that data is being turned into statistical knowledge. It's a theoryless ecorithm, and that's, of course, that's what it's going to do as you throw more compute at it. I think the thing that's surprising is how each time we do it, we expect it to be more intelligent than the last time. That would make sense. But it seems like the leaps are large, larger than we anticipated. Nobody anticipated that ChatGPT 2 to 3 would be such a large leap, that 3 to 4 would be such a large leap, right?

[02:02:28]  Bruce: Nobody anticipated that 2 to 3 would go from writing articles that sound like a human but aren't true to accurately giving you information. Nobody anticipated from 3 to 4 it would act like it had a theory of mind, right? Like the the Alice and Bob example. That's what's surprising about it, that it's more that it's doing something cooler is not surprising.

[02:02:52]  Unidentified speaker: That's exactly what we would have expected.

[02:02:54]  Unidentified speaker: Thank you.

[02:02:54]  Peter: That was helpful. Yeah, I think you've really gotten gotten across just how interesting this whole ChatGPT thing is. Yeah, I would have never completely understood just how weird and amazing this is. So thank you for that, Bruce.

[02:03:14]  Bruce: I do have to say, as someone who became interested in machine learning just as it was starting to become popular and decided to go to school for it. If you had asked me a couple of years ago, is ChatGPT 4 possible, I would have told you absolutely not. That's exactly what I would have said. I would have said, oh, something that you can just go query and you can talk to it and it'll talk to you. It'll automatically learn to use APIs. I would have said, no, that's completely impossible. You cannot do that through statistical theoryless ecorithms alone. It's just impossible.

[02:03:53]  Unidentified speaker: Yeah.

[02:03:55]  Peter: Seems like a lot of people thought it was impossible.

[02:03:57]  Unidentified speaker: Yeah.

[02:03:59]  Bruce: So I was wrong. I think it's surprising what you can do with just statistical knowledge, right? And oh, I know this is where I lost my train of thought. Humans are mostly animals. We are universal explainers, but I suspect that the vast majority of what we call human intelligence is actually just animal intelligence and that we're doing exactly the same thing animals are doing, which is statistical analysis. I think probably everything, what Deutsch calls inexplicit knowledge, okay? That is theoryless ecorithms. That's just statistical knowledge, right? And so an inexplicit knowledge, that's animal knowledge. That's what animals do. And I suspect that to get back to Cameo's question that she asked, of course, most of what humans do is statistical knowledge. And we're not even necessarily better at it than machines, right? This universal explainership where we have this ability to actually think in terms of theoryful ecorithms, that is unique to humans.

[02:05:17]  Bruce: Animals, apes have a very, very limited version of it. And I hesitate to even say that. It just sort of depends on what you mean by explanations, right? I mean, we talked about some of the experiments where the elephants were able to figure out, there's no point in pulling on this rope because my partner is not there and he needs to pull on the rope at the same time. And it wasn't merely, it could not have learned it statistically through trial and error because of the way they set up the experiment. So we know animals have a very, very limited, really only animals with insight. So it's a very small set of animals. It's a very limited form of explanations available to them. And clearly, they're just missing something. They do not have universal explanations. And it was humans, the leap to having language, maybe, I don't know, somehow made this extra leap to where we have this open-ended ability to work with explanations.

[02:06:12]  Bruce: That doesn't change the fact that the vast majority of what we do is statistical.

[02:06:17]  Peter: Are you kind of saying that it could be that ChatGPT is more like a very, very sophisticated animal than a human?

[02:06:26]  Unidentified speaker: Yes.

[02:06:27]  Unidentified speaker: That's an interesting way to put it.

[02:06:29]  Bruce: So I don't think animals could ever do what ChatGPT is doing because they don't have any language ability, for the most part, other than apes, parrots, and a small handful, dolphins, maybe, have super limited language ability. So what you're doing, though, is you're applying something more similar. So this idea of theoryful versus theoryless ecorithms, that's a term that comes from Leslie Valiant, who's a big name in machine learning, you've never heard of him, but you can't go very far in machine learning without having heard of him, because he invented PAC Theory, Probably Approximately Correct Theory. I covered all this in earlier podcasts where I talked about it. We're going to do a podcast on open-endedness next, probably, and it's going to cover, I'm going to read from the book what he says about these things. One of the things that, in my opinion, Deutsch gets wrong, is he tries to make a distinction.

[02:07:33]  Bruce: Sometimes he talks about knowledge in really inconsistent ways. And we're going to do some podcasts about the walking robot, why he says it's not knowledge. He's actually misunderstood the distinction between theoryless ecorithms and theoryful ecorithms. And if his area of study was machine learning, he might have been aware of some of these concepts, but he wasn't. So he was struggling to try to deal with it, and he doesn't have the concepts available. What machine learning is doing is theoryless. It's what I've in the past called heuristical knowledge. It's what animals do. Now, animals' theoryless knowledge is somehow more open-ended than what we do. At least prior to, like, ChatGPT, it was a lot more open-ended. An animal can learn things that are surprising and can deal, because of that, they can deal with life in ways that our robots can't, right?

[02:08:32]  Bruce: It'll be interesting to see if that changes now that we've got something like a ChatGPT to work with, if they're able to take a more general intelligence, for lack of a better term, like ChatGPT and apply it to robots at some point. I know that's like, like Microsoft was trying to release a product called Office 365 Copilot, which was almost like having a Star Trek computer. I'll have to include a link to the show of it's a very cool demo, but like they never released it because they couldn't get it to work quite well enough. They're still going to release it. It's just they're still working on it. But like you can like tell it, I need to make a slide presentation and here's a summary of the notes from the meeting. And it will suddenly open up PowerPoint and put together slides and and then you go back and you modify it. And it's a lot like the Star Trek computer, right, where you can just basically interact with this AI and it's got this general knowledge.

[02:09:29]  Bruce: I think this is a sort of leap that we're talking about, and I'm not quite sure how to define it. And I don't think anybody knows how to define it yet, where it's got this ability to generalize in a way that wasn't possible before. And yet we know it's still just using statistical, theoryless knowledge, just like animals. So I think we're a step closer, I don't know how close, but a step closer to mimicking animal intelligence. And we're still not quite there yet, but like something we've broken the ceiling a little bit, it's started to crack, right? But it's not human intelligence. It's really still animal intelligence that we're talking about. I don't think anything in machine learning, other than maybe explanation-based learning, is really even an attempt at human intelligence. And explanation-based learning really hasn't proven anything well yet.

[02:10:24]  Bruce: Like it just has not developed well. It's not the popular form of machine learning precisely because the results are lackluster at this point. So I think we're still clearly missing something important, right? But most of the machine learning techniques aren't really trying to mimic human intelligence. If you want to look at attempts to mimic human intelligence, probably the best example would be what they've done with theorem provers with propositional logic or first-order logic. Now, obviously, those are nothing like human intelligence. But deductive logic was a very sincere attempt to try to figure out how humans think and to try to codify it. And it ended up being completely missing something. I want to do a class on this where I actually take you through how this all works and how the theorem provers work and things like that. And you can see something's missing, right?

[02:11:24]  Bruce: That you can understand why they thought logic was what humans did to think and why it really isn't.

[02:11:31]  Unidentified speaker: That humans were really doing something else.

[02:11:33]  Bruce: But even back to Aristotle, like when he started inventing propositional logic, he would refer to it not as propositional logic, but he thought he was codifying human thought, right? He was trying to build the first AGI's and failing, right?

[02:11:53]  Peter: That's a crazy way to put it.

[02:11:55]  Bruce: It is, right? They've been working on AGI since back in the Greek times, right? And you can do some amazing things with logic provers and with these things that try to figure out what it can deduce from a set of logical statements. I've actually written some of these in Python because I was wanting to play with them. And it's amazing what you can do with them. Like you can pop in logical statements and it will suddenly say, oh, then I can deduce this. And it's like, wow, that's cool, you know? And it can do it either in a way that is certain or it can do it where it makes guesses. And it can just like sort of try ideas and then if it doesn't find within a certain period of time, assume that there is no proof, you know? And they have different speeds and like there's like tons of cool things that I think really do have relevance to human intelligence, mostly for how they don't work, right?

[02:12:54]  Peter: Well, that this has been a wonderful, Bruce, and Aristotle was the first AI researcher.

[02:13:01]  Bruce: He was. He was the first AI researcher.

[02:13:04]  Peter: I find that very intriguing and interesting way to put it. Yeah, but I've learned a lot and this is that just increases my belief that this is the most wonderful and interesting time to be alive. And I'm really excited to see where this technology goes. So thank you.

[02:13:23]  Unidentified speaker: Thank you.

[02:13:25]  Unidentified speaker: Bye.

[02:13:28]  Bruce: The Theory of Anything podcast could use your help. We have a small but loyal audience and we'd like to get the word out about the podcast to others so others can enjoy it as well. To the best of our knowledge, we're the only podcast that covers all four strands of David Deutsch's philosophy as well as other interesting subjects. If you're enjoying this podcast, please give us a five-star rating on Apple podcasts. This can usually be done right inside your podcast player or you can Google The Theory of Anything podcast Apple or something like that. Some players have their own rating system and giving us a five-star rating on any rating system would be helpful. If you enjoy a particular episode, please consider tweeting about us or linking to us on Facebook or other social media to help get the word out. If you are interested in financially supporting the podcast, we have two ways to do that.

[02:14:14]  Bruce: The first is via our podcast host site, Anchor. Just go to anchor.fm slash four dash strands, F-O-U-R dash S-T-R-A-N-D-S. There's a support button available that allows you to do reoccurring donations. If you want to make a one-time donation, go to our blog, which is fourstrands.org. There is a donation button there that uses PayPal.

[02:14:39]  Unidentified speaker: Thank you.