The Long View: Defining Intelligence

0:21Welcome to A Book With Legs podcast. I'm Cole Smead, CEO and Portfolio Manager here at Smead Capital Management. At our firm, we are readers and we believe in the power of books to help shape informed investors. In this podcast, we speak to great authors about their writings. The late, great Charlie Munger prescribed using multiple mental models and analysis. We analyze their work through the lens of business, markets, and people. Welcome to this episode of A Book With Legs. We're doing something different than we've done typically. Instead of one author and one book, which is our typical style, over 20 times in the year, I wanted to step back.

0:59We've had some questions and conversations that have flourished around similar topics across the episodes, across different professions. And so I want to open those up to kind of ask what I think are some deeper questions. We're going to look across through a computer scientist, a geneticist, a neuroscientist, and a cognitive scientist. I keep asking a version of this same question. Can a human being be fully explained by the machine, by code, by chemistry, biology, by data? Or is there something left over? Today, I want to play you the best answers I've gotten from people who would never share a stage, frankly.

1:35We'll start with the first question and then look at the sharpest place to show up, a small child learning to speak, and then turn and look straight at the machine of our moment of artificial intelligence, three conversations, but all in the same thread. First, Tom Griffiths. He wrote a history of the quest to build a mathematical theory of the mind. Basically, the story of how we got to what we now know as modern AI. I asked him about the one thing these machines still can't do the way a child does. Learn a language from nothing. Here's where he took it. You know, late in the book, you're really talking about the difference between, you know, machine learning or AI and the human.

2:20You're kind of pointing out small differences or big differences, if that matters. So if we could put up slide three, you quote Chomsky here. The child who learns a language has in some sense constructed the grammar for himself on the basis of observation of sentences and non-sentences, i.e. corrections by the verbal community. Study of the actual observability of a speaker to distinguish sentences from non-sentences, detected ambiguities, et cetera, apparently forces us to the conclusion that this grammar is of an extremely complex and abstract character and that the young child has succeeded in carrying out what from the formal point of view at least seems to be a remarkable type of theory construction.

2:56Furthermore, this task is accomplished in an astonishingly short time to a large extent independently of intelligence and in a comparable way by all children. Any theory of learning must cope with these facts. So I thought this was incredible because he's really saying, if you think of how we do machine learning or AI, there's a vast amount of data or intelligence that's provided to it, right back to the idea of there's a store of information as we've learned through computing. And the child doesn't have really much, if any data or store of information to go and execute on. And yet, you know, as Hume might say innately, I and others like Chomsky might say more divinely, are able to do that.

3:44Doesn't that show you how incredibly unique the human is, even while they will use AI to function and learn going forward?

3:56Cole Smead:This is really the biggest gap that I think remains between AI and human minds, right? Which is that humans are able to make these inferences from relatively small amounts of data. Sure. And so if you think about that, it's really a consequence of the fact that the way that neural networks work, they're really good at just learning anything. Right. So you can take your neural network, you can use the same kinds of neural networks to learn all sorts of different kinds of things. Right. The same ones that we use for learning language you can use to learn things about vision and so on. Sure. And so that means that they're sort of like these general purpose learning machines.

4:29Cole Smead:And so they have very weak, what we call inductive bias. They don't really bring much to the problem. Yeah. Right. They can learn all sorts of things, but they need a lot of data to learn it. So if you want to go from like zero to wherever you end up, that whole space is sort of filled up with data. That's how we're going to get a neural network to that point. For the human, right, we know that just that end point, like we only have this much data, right? And so all of the rest of that needs to be what we call inducted bias. And so the humans are bringing to the learning problem some set of sort of, yeah, innate and expectations, but also other kinds of experiences, right?

5:03Cole Smead:The experiences that we have as embodied entities that live in the world, that get to have visual experiences, that get to interact with other humans, that have all of the other things going on that are part of being a human being. All of that adds to the information that we're getting from that limited amount of data that we're learning from. And so that's important in terms of being able to understand why we're able to learn so much from so little, right? There's an evolutionary process that has produced our minds that's filling in that gap that in the AI model is really just being filled in with data.

5:38Cole Smead:So you can kind of think about that big chunk of data, that's standing in for evolution. That's standing in for other forces that have shaped human minds and the other kinds of experiences that humans have. sure um the the other thing about it is that that's something which influences the kinds of solutions that human minds find relative to the kinds of solutions that the ai systems find sure right so part of the reason why ais act in this way which is often counterintuitive to us and do weird things and have these weird failure modes and you're you know like they succeed at one thing and then fail at something that's right next to it yeah it's called jagged intelligence that is a consequence of not having the same inductive biases as of us right they have they're good at the things that are in their training data.

6:18Cole Smead:They're less good at things that aren't in the training data. We don't know what's in their training data. They learn weird idiosyncratic patterns because they're just sort of learning to mimic behavior that they see in their training data. And so when we learn something, we're guided by those inductive biases and you can make pretty good inferences about what another human being will be able to know or do based on things that they know or do. But we can't make those kinds of inferences for AI systems because they're operating with just like a completely different set of inductive biases. Yeah.

6:43And I mentioned to you before, it reminds me of like a new intern to an industry, right? It's got jagged intelligence or information. It can be refined over time, but you may, might do something that's helpful. And then you're thinking like, well, what do you, what did you just do? I don't understand what you've given me. Um, and makes it very human, like in that respect in many ways. Um, I want to go, uh, Oh, one other thing I want to mention on this totally, I mentioned, you know, this technology conference, uh, Cosm, and there was a gentleman when, when I was thinking about, you know, Chomsky's idea, I was at this conference, and this gentleman made a presentation on brains where they have epileptic attacks that are regular, and how do you solve for that?

7:22And so he mentioned that what they do, what neurologists do is they sever the hemispheres of the brain, because it's the misfiring between the two hemispheres that's causing these epileptic issues. And so we know that physically, once the two hemispheres are separated, they can't talk to each other. And you obviously control one part of your body with one hemisphere and another part of your body with the other hemisphere. And we know also that each hemisphere has different things that they take data in and process based on. And his interesting presentation was, based on this, if you're thinking in your left brain, you should not be using your left hand to respond, because your left brain would correspond with your right side of your body or your right hand.

8:04And what they found in studies is when people have severed brains, they're just as likely to use their left or their right hand to respond. And so the question is like, how is this going on? Kind of like, how is the child able to develop this innate ability? And he argues that there is something divine going on. He said it's a soul. That's the divine nature. And when Chomsky said that, it played along that idea of there's something more divine. And who knows how you define divine, but there's something more incredible or awesome that is fairly unexplainable. and to your point, tough to replicate and compute.

8:40Cole Smead:Yeah, I mean, I think that's an interesting way of framing it in terms of like the, it kind of goes back past Chomsky, back to Plato. So Plato had the same answer to this question, right? So Plato was trying to explain how it is that we come to know the things that we know from limited experience in the world. And the way that he answered that question was that we're not learning things, we're remembering things. And we're remembering things that are like stored in our souls that come from the experience that we had in some platonic realm, right? And comes through. And so I think Chomsky would think about this in much more materialistic terms, right?

9:18Cole Smead:That he really thinks that it is biology, right? It's, you know, the 21st century scientific analog of the soul is the genes or something like that, right? And it's like built into our genetic heritage. So Griffiths looks at the child and sees something almost divine. Now here's a neuroscientist who builds a neural network for a living, Gaurav Suri, coming at the same leftover thing from a completely different direction. I think that human behavior is a beautiful emergent product of many things. It's not this reasoned self-interest alone. It's not we are capable of empathy. We are capable of well-spirited gambles.

10:11We're capable of proceeding without knowledge. We're capable of happiness even around suffering. Machines are not having these aspects of our experience. and we are capable of, even based on our experience, we're capable of persisting in adversity. We have many aspects of intelligence. The LLM is based on one of those aspects, one of many. And the LLM, meantime, it's based on language pattern prediction. And that is one basis of our intelligence. But by far, it's not the only basis, right? We're not limited to words. Thoughts exist without words. Thoughts can exist in images. So many times I'm listening to somebody and I already know I think they're wrong without having any words for why I think they're wrong.

11:03Or I'm watching Jeopardy and I know I know the answer without the answer coming to mind, right? So this happens a lot, that's because language is not thought. And we have other modes of thought. And the large language model deals in the modality of language. We have goals. The large language model doesn't have any of its goals. We have efficient learning from a few instances. We have the capability to learn very quickly. Large language models have to be trained billions and billions of times. There are many aspects of the human experience that are outside the large language models. I do want to say, and this goes to an initial conversation, I see in principle, there isn't some reason why we can't understand our working.

11:52Our workings are beautiful, profound, sublime, and understandable. And it is in this understanding that I find my awe. faith gambles love the parts the machine doesn't have you know the really human part that we think about our own families or our spouse or our kids now listen to a geneticist katherine page hardin who spent her days reading people's dna and really the coding of that dna you expect the hardest determinist in the room instead she found something that sounds a lot like grace

12:26Cole Smead:you know i think the thread that you're picking up on here is that the line between the individual and their community is is much fuzzier than we'd like to imagine none of us is an island none of us is a fortress and that means our individual choices reverberate and have implications for other people and and they're structured by other people's choices in this way where you know the inside of our body and the outside of our body is a lot less clearly demarked than we would imagine I write about this when I talk about you know there's a so-called nature nurture debate and if you ask any scientists they'll be like the nature nurture debate is dead like it's always nature and nurture in combination and that also even the division between nature and nurture is really artificial because your nurture changes your body.

13:24Cole Smead:It changes the expression of your genes and your nature by shaping your behavior is then providing that nurture to other people, right? And I think that example you just gave of if you have a man who is predisposed towards antisocial behavior because of his genes, but then he's passing on both those genes and the environment that he is shaping to his children. Yeah. It's nature-nurture. It's one combination. I think the other thing is just as a geneticist, I see that all of us are just a few genetic mutations away from being profoundly different people with different capacities and different – it is just by the luck or the grace of whatever, that you didn't have that experience.

14:16Cole Smead:And also that DNA strand that runs through them is 99 % 99.5 % the same as mine. And so I really think genetics has given me this profound appreciation for no one is a monster. People can behave monstrously, but they're doing that in part because of genes and environments of natures and nurtures that run through all of our experience. Like I have mostly the same DNA as someone who's, who has committed the most atrocious act that you can think of. And I, that goes a long way to undercut that, both the separation between the individual and the community, but also the separation we might feel from someone who's behaving in this really abhorrent way, we are connected in the same tree of life to that person.

15:21And last, Charles Murray, a social scientist who I had a ton of fun visiting with, spent most of his life as a contented agnostic, a happy agnostic is what he termed himself, and came to faith, this idea of faith, almost against his will and really late in life alongside of his wife. He has a name for the leftover thing. You make the point, and I've heard this term used a lot. People talk about God-sized holes, and you'll hear people explain that in their own personal experience. But you use it differently in your book. You say that in your eyes, modernity has hidden God-sized holes. In other words, it's the thought, it's the cultural view that's covering that up.

16:08Can you explain that? Well, I think it has to do with, Bill, your father, not getting serious about religion until he had serious cancer.

16:21Cole Smead:We, in the 21st century, and basically in the 20th century as well, sort of assume we've got 80, 90 years coming. I mean, odds are we're going to live that long. We're probably going to live pretty healthy lives. Yeah, we know people who don't do that, but there is kind of an assumption we're going to be okay. Just the same as in more time, soldiers assume I'm going to be okay even though everybody else around me dies. And in doing that, you are enabled to put off all sorts of things that you're more comfortable putting off, such as the prospect of death, such as taking the serious questions about human existence and the human condition seriously.

17:08Cole Smead:And in addition to being protected from a lot of the miseries that people had before the 19th or 20th century, their children dying very commonly, a spouse dying very commonly, terrible diseases that left you helpless and couldn't be cured. We've also developed in the 20th century the ability to amuse ourselves constantly. And that's grown by leaps and bounds in the 21st century, where you can entertain yourself 24-7 with no problem at all. Correct. Right. So in all of those situations, you can kind of shove the big questions out of your mind. And a crisis such as a health crisis, such as the death of a loved one, is one of the few ways remaining in which you are driven to the depths of despair.

18:01Cole Smead:And in that depths of despair, you discover you do have a God-sized hole after all. You're not making it up. The hole was there all along, but you were never forced to recognize it before. I didn't hear it until the mid-1990s. and I heard it from Charles Krauthammer, the late columnist. Sure. And it just hit me. Why is there something rather than nothing? And I had no ready answer to that because I am a good Aristotelian, or I was even when I wasn't religious, and things have to have a creator. This had to come from somewhere. Well, where did it come from? and one way out of that, of course, is that you could say, oh, if you try to answer that question, you just enter an infinite regress.

18:56Cole Smead:You know, it's God created it. Okay, who created God, et cetera, et cetera. Well, that doesn't deny that there's a mystery here. It just means that maybe it's a mystery with a capital M instead of a small M. And that's what I came to believe. And that was one nudge that kind of pushed me off dead center in my attitude toward religion. And it said to me, you've got to take the idea of a creator seriously. That was a big step forward. At the same time, something else crossed my mind that wasn't as powerful, but it's the regularity of mathematics, the simplicity of mathematics, where you have very profound around physical phenomena that can be explained by laws that give you exact results from very simple equations.

19:50E equals mc squared is the most famous, but Newton's law of motion,

19:57Cole Smead:they're also very simple. Same is true of many others. Mathematics is way too elegant to have been produced by chance. And that leads you to say, well, if it wasn't produced by chance, what produced it? Four people, four completely different fields, code, neurons, DNA, statistics, and every one of them pointed past the machine to the same missing piece. They called it different things, a soul, emergence, grace, a God-sized hole. You can call it whatever you want, but not one of them could make the human being add up without it. Keep that thread in mind because we're about to watch it show up in the single sharpest place it appears anywhere on this show.

20:45Hi, I'm Cole Smead, CEO and Portfolio Manager here at Smead Capital Management and host of this podcast. If you enjoy this podcast, I'd like to invite you to check out SmeadCap.com. At our firm, we are stock market investors. We advise investors who play the long game with a discipline that has proven success over long periods of time. Learn more about our funds at SmeadCap.com. Past performance is not indicative of future results. Investing involves risks, including loss of principle. Please refer to the prospectus for important information about the investment company, including objectives, risks, charges, and expenses.

21:20Read and consider it carefully before investing. Smead Funds distributed by Smead Funds Distributors, LLC, not affiliated. All right, let's slow it down. And let's look at the one place that question gets really concrete. The place that stopped being philosophy and starts being something you can almost touch. It's a small child learning to speak. I have four kids myself, so I think about this as they're growing up and their brains are learning to learn, in effect. Nobody programs the child. It's working from almost nothing. A few years of messy, half-heard conversations. We think of babbling. We think of the weird words that kids use in lieu of the words that we would.

21:58And what they pull off is something – maybe our largest, most expensive machines have trouble to even match. We think of all the progress of technology to get where we've been, and they have matched that linguistically, but yet it took data and energy and time and money. Our children do require some of that, but not to the same level. One man built an entire career insisting that proving something is wired into us from birth. Here's that argument and the rebuttal. I'm really blown away that when Noam Tronsky asked a question 70 years ago, it's still reverberating, I would argue, and it's something that's really affected me personally.

22:43A small child learns to speak from a tiny, messy sample of language. Effectively, they have a couple of parents framing that and others around them in a stunningly short window. And every child does it the same way, regardless of how smart they are. Intelligence is not the main factor. Chomsky said no system that just learns word by word could pull that off. In other words, it's not just the words themselves. It is the phrase and it's the formatting of language and how we communicate with each other. Something that has to be built in. Today, our biggest AI models learn language from something like 50 ,000 years worth of speech.

23:17A toddler does it on a rounding error of that. So who's right about what language and the mind actually is? Here's the debate in three voices. Tom Griffith lays out Chomsky's actual argument, why a simple word association model of language was never going to work. We're kind of back to the Wilkins idea, if you will, which Chomsky tried to tackle. What was novel about Chomsky's work at that time?

23:43Cole Smead:Chomsky was a young linguist who had a very different view of what linguistics was from all of the other linguists at the time. So for a lot of the other linguists at the time, what it meant to be a linguist was you were going out and you were collecting information about what languages are like. And then maybe thinking about what some of the correspondences were across these languages, what some of the general principles were that characterized human languages. Sure. Sure. And Chomsky wanted something which was different from that. He wanted to have a way of characterizing, you know, if you had a single language, could you come up with something like a mathematical system that would allow you to characterize all of the sentences that were grammatical statements in that language, right?

24:23Cole Smead:So trying to create a formal system, right, that gives you a way of producing all the sentences that are in that language and nothing else. Okay. And so this approach, which is called generative grammar, led him to then ask this question, what kinds of formal systems do you need in order to capture the structure of human languages? And so he was able to show that some simple kinds of systems that were like things that the behaviorists and the information theorists had assumed would be sufficient to capture language weren't going to work. And in fact, you needed something much richer that had a sort of like internal structure, right?

24:58Cole Smead:Like ideas, like noun phrases and verb phrases and so on, that then could capture the structure that sentences have in order to get even close to the structure of English. Sure. Explain what's a finite state language versus something that's not a finite state language. So this is the distinction that Chomsky was focused on in terms of ruling out a sort of simple model. So a finite state language, you could think about it, again, going back to our board game analogy, as a language that you could generate just by creating a board game where a bunch of positions that you're playing pieces can be in.

25:35Cole Smead:And you have moves that can take you from one position to another. And then as you move into each position, you produce a word. So I go from the start position to the next position. I produce the word the, I go to another position, I produce the word dog and so on. And then maybe I go from that position to the end of the game and I produce the word runs. And now I produced a sentence, the dog runs. That structure allows you to capture certain things about language, right? You can imagine now you can capture the fact that you can have sentences like the dog runs and the dogs run and you're reusing parts of the board, right?

26:15Cole Smead:When you're producing those sentences. There's not a unique trajectory that you have to follow for every sentence. And so it allows you to capture some of the structure that exists in language. And for the behaviorists, that was nice because it suggested that you could learn language just by learning associations between words. So you could learn that after dog, then you're more likely to produce the word runs, right? And after dogs, you're more likely to produce the word run. And, you know, you could sort of learn these associations between things. And it was good for the information theorists because they wanted to be able to estimate the probability of different pairs of words occurring next to one another.