Into the Impossible With Brian Keating - Noam Chomsky on AI, Neural Networks, and the Future of Linguistics (#349)

Episode Date: September 20, 2023

Noam Chomsky is one of the most influential and highly cited scholars of our time. He is a pioneer in the fields of linguistics and cognitive science. He has deep thoughts about communicating with aliens, meditation, Elon Musk, and free speech. Chomsky is a prolific author and known political activist. We avoided politics, as is the custom for my interviews. Our conversation also covers the Turing Test, neural nets, and artificial intelligence, including why he expects Elon Musk’s Neuralink project to fail. Key Takeaways:  Introduction (00:00) Does Noam Chomsky believe in extraterrestrial life & could we communicate with E.T.? (09:01) Busting (or confirming) linguistic myths with the master (23:57) “There is no scientific method. It’s just being intelligent.” (39:19) Artificial intelligence applications in cognitive science (58:13) Chomsky University would encourage discovery (1:09:46) Thoughts on the negative reaction to the Harper’s Magazine article (1:21:52) What object or knowledge would Chomsky put in or on his monolith? (1:29:59) What did Chomsky think was impossible until he did it? (1:32:16) — Additional resources:  Noam Chomsky has been called "the father of modern linguistics." He has been a professor at MIT since 1955 (now emeritus) and continues to teach at the University of Arizona at the age of 91. Chomsky has received numerous awards and honorary doctorates, along with being a member of multiple professional societies. He has written over 100 books covering topics including linguistics, politics, and philosophy. Read “A Letter on Justice and Open Debate,” the letter in Harper’s Magazine which Chomsky signed https://harpers.org/a-letter-on-justice-and-open-debate/ Find Chomsky on the web: https://chomsky.info and Facebook: https://www.facebook.com/Noam-Chomsky-294468630182/ 🥗 Thanks, HelloFresh! Go to HelloFresh.com/50impossible and use code 50impossible for 50% off plus 15% off the next 2 months. 📝 With a MasterClass annual membership, you can take one-on-one classes from the world’s best for $10 a month with your annual membership, get unlimited access to every class — and even better, right now, as an Into The Impossible listener, you can get 15% off when you go to MASTERCLASS.com/impossible. 🧑‍💻 Visit LinkedIn.com/IMPOSSIBLE to post your job for free! 🎤 Join me and ⁦Lawrence Krauss for an Onstage Dialogue ⁦at the San Diego Air & Space Museum Tuesday, Oct 17, 2023 at 7:00 PM: https://www.eventbrite.com/e/live-onstage-dialogue-brian-keating-lawrence-m-krauss-tickets-699430514497 ➡️ Follow me on your fav platforms: ✖️ Twitter: https://twitter.com/DrBrianKeating  🔔 YouTube: https://www.youtube.com/DrBrianKeating?sub_confirmation=1  📝 Join my mailing list: https://briankeating.com/mailing_list  ✍️ Check out my blog: https://briankeating.com/blog.php  🎙️ Follow my podcast: https://briankeating.com/podcast  — Into the Impossible with Brian Keating is a podcast dedicated to all those who want to explore the universe within and beyond the known. Make sure to follow so you never miss an episode! Learn more about your ad choices. Visit megaphone.fm/adchoices

The neural net systems are just the wrong place to look. They don't have the kind of architecture, which is involved in thinking. We have to find something else. It might turn out to be at the molecular level, with the level of RNA. Molecular level, you're really getting massive possibilities of computation. So maybe just duplicating a neural net will tell you nothing, because you're not looking in the right place. We don't know.

I mean, the thing to do is do the science first, then worry about the engineering. Any sufficiently advanced technology is indistinguishable from magic. Open the pod bay doors now. We've never had a linguist on, and I wonder if it's possible for you to give a brief overview of why linguistics is important, and what inspired you many years ago to pursue it. It's been understood for millennia, actually, that since classical Greece, that possession of language is the almost the defining characteristic of human beings.

It's a species property. It's common to all humans. It was assumed for centuries that it's common to all humans. Now we have good evidence for it. There doesn't seem to be any group variation. It's apparently a very recent development last couple hundred, maybe 200,000 years ago, roughly about that,

nothing in evolutionary time. It has, its properties have no analog in the animal world, essentially none. So it's a true species property, common to humans, distinctive of humans. And it's the core of our creative capacities,

the reasons why people got Nobel Prizes are thinking it's just central to all of our lives. if you follow the Elphic Oracle, let's say, I don't assume that our goal is know thyself, then this is the place to look. And in the context of linguistics, I have a couple of things. Again, I beg your forbearance as we proceed to things that are possibly of interest to me, and me alone, but these will be related to my role as a father, of many young children and some of the constructs are hypotheses that I've tried to test

throughout unwittingly using my children as experimental research subjects. But perhaps we'll get to those in a little bit. I think what was so interesting to me is that you drew a connection in your 2017 talk between consciousness and linguistics. And I wonder, first of all, is it, is it possible, in a sense, to link a mathematical construct of what is linguism or what is linguistically a statement that can be proven? Because it seems to me that there are vast relevancies between linguistics and mathematics, and that's no accident. You've contributed a lot to the quantitative interpretation of it.

But let me flesh out what I'm trying to explain. In mathematics, according to GERDL, it's possible to prove that there is an incompleteness in mathematical formal logic that there'll be statements that cannot be proven consistently within the system of formal mathematics itself. Is there an analog of Girdle's incompleteness theorem in linguistics? Is there a formal system that can define what is outside the bounds of linguistics, or is it just a combination of neurological, motor skills, etc. How is, is there a relevant or is there an analog of girdle's theorem for linguistics? It's not really a girdle's theorem, which only applies in very specifically defined formal systems.

Language is not a formal system. So, of course, it doesn't apply. But in fact, in California once they spent some time without. Alfred Tarski, and he couldn't understand why anybody understands language because you can formulate the logic, but the liar's paradox. You know, so you're interested in it. But it's an organic system. You're interested in its properties. You can immediately find things that aren't in it by looking at its properties. Take a look at the visual system, begin to understand it. You can find things that aren't going to be accommodating. by the visual system, but has nothing to the girl's theorem. That's just that's what happens when you look at any physical entity,

and the language system encoded in the brain is just one physical entity. So we can find, we can search for its properties, so we will find things that are outside those properties. In fact, there's very interesting work on this, incidentally, even neuroscientific, neurolinguistic work. So, for example, you can construct formal invented languages which violate the principles of the universal principles of human language. And then you can ask what happens when people are exposed to these insistence. And you get some interesting results.

So, for example, one of the deepest and in some ways most surprising properties of language is that the rule of the core rules of language, the part of language that is concerned with basically constructing thoughts. You can divide the language system and basically two parts. One part is concerned with constructing linguistic expressions, which are expressions of thought. The other part is externalizing it to some sensory motor system. It's kind of like the internal program and your laptop and the printer you attach it to, you can attach it to one or another printer. And in fact, the internal system of language, we happen to be using speech.

But if we were deaf, we would be using sign. And it's essentially comparable as just a different printer. It's the same internal system. So if we keep the internal system, the core of language, turns out it, pays no attention to things like linear order, only pays attention to the structure of expressions. Which has a very funny consequence. It means that your children, for example, when they're acquiring language, pay no attention to 100% of what they hear and only pay attention to what they never hear. They hear things in linear order, but the rules that they use pay attention to structure,

which they don't hear, they construct it in their minds. Now this principle called structure dependence allows you right away to develop impossible languages, namely languages that use linear order. So for example, you could construct a language which is like ordinary language except the way it uses negation, say, is not the way ordinary language is do by structure. by structural positions, but rather by linear order.

So suppose you invent a language in which, if you want to negate a sentence, the third word will present the sentence will be not. That's a trivial problem to solve. But if you give it to humans in invented languages, turns out that they can solve the problem as a puzzle, but the language areas of the brain. are not the ones that are activated. Rather, you get diffuse activation as for puzzle solving. If you give them an invented language, which keeps to the rules of ordinary language, then they can learn it but the language areas at work. So there you can find,

you can indeed study impossible languages and the way the brain reacts to them. And this is one of the most striking examples. One thing I wanted to discuss maybe segues into it, and I'm sorry to use up my perhaps one request that your forbearance so soon in our conversation known, but I was thinking about the communication problem, the so-called communicating with extraterrestrial intelligence. And I want to understand,

if you have a perspective on this, because as you make the case so often, a lot of language, as you're saying, is not even the verbal or structural or even the linear processes by which language can be acquired by human beings. And yet, if you imagine the problem of communicating with an extraterrestrial civilization, purely using a printer, or using binary code or some such form of flashing lights converted into Mars code, perhaps. Is it, and assuming the intelligence could decipher such a thing, first of all, assuming they exist? By the way, do you believe that there are extraterrestrial civilizations?

I promise this won't be a long... What's that? Fermi paradox. Yes. Do you believe that they exist, or do you have a reason to suspect that they might not? Intelligent communicating technological intelligence? They may. In fact, they may be all up there.

If they have any intelligence and they pay any attention to what's going on on Earth, they'll get far away. I don't do with people with creatures like us. That's right. I think that's probably the answer to the family paradox, as a joke. But there may be actual intelligence. Maybe not. And then assuming that... Notice intelligence is a very rare property.

you may have seen Ernst Meyer's article on this responding to Carl Sagan seen that discussion interesting discussion so Carl Sagan's from the point of view

of an astrophysicist there's got to be all these planets out there very much like ours it's got to be like this at Meyer from the point of view of a biologist he was a grand old man of American biologists as well

maybe, but he says we have one case that we know about Earth. There have been about 50 billion species. Some of them are biologically successful. That means they survive in proliferate. Some of them are not. The ones that survive in proliferate are the dumb ones. Things like bacteria mutate very quickly, no problems. or say, beetles.

You remember, Old Dan's famous comment that God must have loved beetles. Made so many of them, right? They're fine. They find a niche. They just stick there and dig it by. But as you go up the scale of what we call

intelligence, survival gets lower and lower. Large mammals, for example, are very rare. The only reason there's a lot of cows is because we domesticate them. But if you look at the wild, in the wild, say, apes,

you know, very few, they don't survive very well. And if you take humans, it's probably only the last couple hundred thousand years, which means that several billion years of life went by and never know

humans, nothing with what we call higher intelligence. That's right. Extrapolate and ask what might happen on other planets, the chances of developing higher intelligence might not be very high. Yeah. It just might not survive. And even if they're intelligent, it doesn't mean that they're technologically advanced,

that they're able to interpret and build devices. You know, I always say it takes, you can't build a solar panel. You know, solar panels weren't built using solar panels to provide power. In other words, there's a hierarchy of energy scales that were needed to construct something as sophisticated or not as a photovoltaic sound. Tiny percentage of human life. But let's suppose that there's something like human intelligence there. Would there be ways to communicate?

Well, I think the thing to do is not to look at the printer. No, they could have used, just like humans, we can use one or another sensory motor system. We can use our hands. We can use touch. You can even learn language from touch. Yes. You can learn. We use sound.

because it's convenient to use some other. So I don't really think that's the issue. The issue is the internal system, a system that constructs infinitely many thoughts, basically. And if you look at that, there's a good reason to think that there might be a mode of interaction, namely arithmetic. If you'll take a look at the structure of language,

it's the internal system. There's pretty good evidence by now that it's based on the most elementary computational device, namely binary set formation. And if you take a look at binary set from that, you can construct the infinitely many structures and so on. And if you take a look at that, you can, from binary set formation, with a lexicon of one element, you basically get the basis for arithmetic. is kind of another direction from which you could look at this. Marv Minsky, you know,

I found a guy about 50 or 60 years ago, did some experiments in which he just took the simplest touring machines, fewest states and symbols, and asked what happens if you just let them run wild. turns out almost all of them crash. Either they get into an infinite cycle or they terminate. But some survive, and turned out they all had the successor function. And then he concludes, well, suppose evolution is getting to the point where it's developing systems that have some of the capacities of touring machines.

it's going to hit on the simplest things and the simplest things will give you something like the basis for arithmetic and maybe that gives you language that's a point where there's possible convergence in order to pursue it you have to show that so for example this point I made about structure dependence that actually follows directly from the fact that the basic computational

system for languages by a reset formation, because that does not yield linear order. So if that's the system that's in your child's brain, it's never going to use linear order. It'll use it for communication, but that's because of the sensory motor system. The sensory motor system requires linear order. We can't talk in structures, so you have to linearize the thing. But that's the property that printer has basically nothing to do with language. Sensory motor system was in place hundreds of thousands, millions of years before language emerged. And it's basically nothing to do with it, just as your printer has nothing to do with the program in the laptop.

You said this place was steps from the water. We just haven't found the steps yet. How much did we save? Enough. Enough to get lost! Or you could book a stay with Hilton. Welcome to your oceanfront room. Just steps from the water.

The Hilton sale is on now. Book on Hilton.com or the Hilton app and save up to 20% to get the stay you expected. When you want savings, not surprises. It matters where you stay. Hilton for the stay. So in the absence of all systems, though, so the absence of a neuromechanistic touch sound

with these aliens, merely communicating only with arithmetic, you know, symbolic symbolic logic, that would be sufficient? I assume that they have some mode of externalizing what's in their heads. If we can latch on to that mode of communication, that printer that they're using, then we could go back to the internal system. good reason believe that they would have the successor function in addition. We have the successor function in addition. It's part of our language. So maybe that could be an entry point.

Interesting. So you mentioned artificial intelligence. Hopefully we'll get around to that in a little bit. Are there subject models, does linguistics benefit from, as I know, colleagues here at UC San Diego have studied consciousness and sensory perception in subjects that have had damage to their brains and the problems that they illustrate have revealed patterns of understanding of how actually the brain works. And I'm wondering, is that the case in linguistics, too, are there deficiencies in subjects that from which you learn more about how those of us

who are blessed not to have deficiencies in that realm, how we actually process language? Well, there's quite a lot of work on language pathologies, deficiencies of one sort. Its work actually began about 50 years ago with some classic work by Eric Lennelberg, an old friend of mine who founded the modern biology of language. and there's a great deal of recent work on it. So let's take one example. There's a well-studied case of a subject called his name. The name that's given to him is Chris.

He's a young man who has extremely limited cognitive capacities. He can do almost nothing. But he has amazing linguistic capacities. Expose him to a language, learns it very quickly. He's mastered dozens of languages, and he's kind of obsessed with it. All he wants to do is learn another language.

Actually, they tried this test with him that I mentioned before about impossible languages. So the language that you give, and they tried the same test with Chris. Give him an invented language you've never seen before, or even an invented language that keeps to the linguistic principles, learns it very quickly. Give him a system which uses something like linear order for negation

or other violations of structure dependence. Total blank. You can't make any progress because he can't solve puzzles. And there are lots of different kinds of cases. There are cases of subjects who have almost no cortex, tiny amount of cortex, but complete language capacity. Oh, there, you know, I mean, this actually began with the study of aphasia back in the early 19th century. But since then, there's been, especially since Lennelberg's were extensive study of a range of different language pathology. and the effects that they have.

So I want to get into some other speculations, perhaps, not as well-founded as some of the other topics we'll get into when we come back to your talk on consciousness. When, is it true that Richard Feynman used to say that he was surprised when he realized that not everybody counted the way that he counted? In other words, I believe he would count and he would, hear in his mind the numbers, one, two, three, et cetera. And he realized that that was, you know, that was just the way that he visualized or heard numbers, if you will, and that's the way that

he counted. He realized that some of his colleagues in MIT or, you know, fellow students, that they would see numbers moving across their mind. Is there, is there, first of all, there, there is this continuing running monologue that I always say, you know, if somebody heard my inner monologue, you know, they might think I'm saying, you know, if they didn't hear it, but they certainly wouldn't feel I'm saying, you know, just the endless bombardment of language. Is it, is it the case, though, as with Feynman's, you know, finding, do people, some people see words or, you know, when they're having these internal monologues you've spoken about in the past, or do they merely hear them kind of the way I do? Is that something that's universal,

or is it, is it, you know, bimodal and some people hear it, some people see it. How do people... Let me ask you a question. When you're typing a letter, not paying a lot of attention, just typing, not paying much attention, do you ever notice that you make typographical errors where you type a word that sounds the same way? Yeah.

Like, you suppose you're planning to say write W-R-I-T-E, that you write R-I-G-H-T. Sometimes, yeah. Does that mean I'm crazy, or just be committed? It happens to me, too. I think what it means is you're hearing the things. I think you're writing, but you're actually hearing. And that writing is a kind of very peripheral activity.

Hearing is much deeper embedded. The writing is just, it's even more remote than the printer. It's a way of mapping the printer into something else. secondary printer. So actually when you are doing something like typing, you're often just hearing yourself. And that's why you make mistakes like that.

Are there differences in the way people do this? As far as I know, it hasn't been investigated much as far away out of the periphery. I mean, the deep questions are, you know, what's going on with the kind of things we were talking about before? Yeah.

If you have certain kind of brain injury, what's that going to do to your language fact. Here, there's lots of work. Are there any analogs in, you know, Ramachandran's brief tour of human consciousness? He speaks a lot about, you know, synesthesia and kind of pathological circumstances, brain injuries,

people will, you know, smell the color orange or things like that. Have the linguistic differences between such patients been studied? In other words, you know, how does it affect their ability to do just what we said, you know, typing, writing, hearing, seeing? At the level of typing, I've never seen anything. But the level of speech errors, so cold, there's quite a lot of work. Actually, Susan Curtis and your university is one of the leading specialists on this chadesteper. Yeah, I'd definitely like to have her on the podcast as well.

So getting just a couple of more language, you know, maybe popular myths or whatever, As I said, I have many children, and I'm very fortunate to have so many. But I heard it once said by a mathematician that he made sure to speak several thousand words a day to his children. And that he believed that there had been some studies that showed that children begin to speak only after they've heard a million words. So that would be a year's worth of 3,000 words a day roughly. So you're shaking your head, so I guess almost nothing good. Actually, what the studies show is that children don't pay much attention to what their parents are saying. They pay against you to the ambient environment.

But you can ask yourself, I don't know, your background, but take, say, my background. My parents were immigrants. So they knew English, personally well, but with accents. I don't speak like them. I talk like the kids on the street. That's my good dialectician. could figure out quickly that I come from northeast Philadelphia, not from the Ukraine. Right.

And that's pretty, nobody understands why, but children usually pick up the dialect of their peers. And not really, I mean, the parents try to train them, and the kid may listen, but then goes back to its ordinary behavior. the very little impact of parents' efforts. It was at one point a belief in the child language literature that there's something that used to be called motheries. Mothers talk to children in very special ways and not supposed to help them learn. But as the careful studies took place, it turned out,

the kids just weren't paying attention. So there's another myth that children start learning language in utero. I guess from what you just said, that is true. Well, they learn something. What they learn in utero is probably prosody, no, pitching. Yep, density. But the experimental evidence is that shows,

originally work of Jacques-Mille, the French cognitive scientist, you take a two-day-old child, about as early as you can begin to test anything. And the tests usually have to do with intensity of sucking. It's about the only thing you can measure. You get surprise reactions if the intensity of sucking increases. It means the infant is interested.

So you can distinguish things that cause surprise from those of done. And just using that experimental technique, He was able to show that a two-day-old infant can distinguish the language of its mother from the same language spoken by a perfectly different language spoken by a bilingual woman who speaks both that language and the second language. the infant could distinguish the voice he's never heard talking his mother's language and talking some different language

and since then there's been that shows something's going on in utero and since then there's been quite a lot of study of it and it turns out it's not the actual language it's a certain category of languages ones that have similar prosthetic structure this has even thought to the point of people putting sound devices on the uterus of a cow,

seeing what you can hear, and you hear a sort of muffled speech. So you can, something's coming through that the infant is picking up something or other. The point is that children are pre-programmed to acquire language. It's a very striking fact that, I mean, an ape, for example, a chimpanzee has about the same auditory systems, humans. And if you give in an ape and an infant exactly the same environment, the infant

immediately hear speech and picks up language in a regular fashion, almost reflexively. The ape just hears noise. It doesn't matter how much training you do, can't do anything. So it's just an internal, your genetically pro, the child is genetically programmed to pick up all the noise in the environment and say, I'm only looking at this. And as I said, it's kind of striking that, in fact, an infant doesn't pay attention to 100% of what it hears linear order, pays attention to what it never hears the structure that it in its brain construct, which is a pretty dramatic finding. Are there languages, this is the last of the myths that I'm going to ask you to bust or, I don't know what the opposite is, but bolster. I had a phenomenal professor of high-energy particle physics when I was a graduate student at Brown University since deceased, Kyeongsokang, and he used to tell us, in a kind of a mercurial smile on his face, he used to say Korean is the most logical language, because the pictograms, the glyphs of the language were, in some sense, reminiscent of the facial motor system that was to be employed in the position

of the tongue, et cetera. Is that a myth or does that have any validity? There's some truth that it's not about language. It's about the writings. The writings, yes, right. The writing system is very peripheral to language. Writings are very recent phenomenon under a very small part of the human population until recently. It's true that the Korean writing system, and it does not so much facial expressions, but it does reflect phonological properties to an unusual respect, an unusual extent to the green level.

So there's something to it, but it has essentially nothing to with language. Right, that's pure writing, yeah. But I guess in the sense of the written language as a tool for acquisition of verbal language, I mean, there's some at least peripheral knowledge. If I can read the language and I can write in the language, it may assist in some of the breakdown of some of the cognates, perhaps, especially if you translate from one language to another. But notice that that's very late in language.

Oh, yes, sure, absolutely. A two or three-year-old has an informant's language capacity. Yeah, and that... are exhibited. So, for example, when an infant is in what's called the two-word stage, it just says two words, you know, me hungry or something, whatever. No, yeah. More. But at that stage, the child is understanding much more complex sentences. You can show that by trying to introduce errors into the more complex sense. sentences, the kid can't understand it.

In fact, there have been studies in which this is called telegraphic speech. You know, you're talking just none of the small words, just nouns and verbs. So you give a kid who's in the telegraphic speech stage three conditions. One normal speech, one, the child's own telegraphic speech, the third. telegraphic speech with the small words randomly introduced. So three different conditions. Turns out the kid can't understand its own speech. Can't understand the randomly distributed ones,

but it can't understand normal speech. Because what's going on in the head is much beyond what's coming out of the printer. And so maybe there was one more. one more comment that I had that was sort of related to popularized myths perhaps rather than fundamental mysteries and I'm trying to find in my notes.

Oh yeah, here it is. So we have in language, of course it's written but in this particular context, but there is a connection to the spoken language as well. There's a sense that linguistics, If it is to be a hard discipline, a regularized discipline with the rules, et cetera, how is that consistent with the fact that language, at least in the realm of vocabulary, changes regularly.

I'm sure I was dismayed a few weeks ago to learn that Miriam Webster's dictionary now no longer marks as incorrect spelling, the word irregardless, which was always the bane of whenever when my students would use such a word in spoken or written. language, they now accept it. In fact, you can type it into your computer and my computer will not flag it as making a spelling error or even a grammatical error. If we add to the vocabulary, why is it, is it not impossible to imagine that even grammatical things such as subject verb agreement, is that a slippery, you know, shifting language zeitgeist? Or are there certain laws that are immutable, almost laws of nature when it comes to language?

If you don't understand something, it looks as if it varies all over the place. So if you go back, say, 60 years, biologists believe that organisms were so diverse that you have to study each one on and so on. Nothing to say about them. By now we know that they're so uniform that there's even serious thought about the possibility of a universal genome. As if all organisms are basically the same, ones that came out of the Cambrian explosions, have very few different life forms, very small number. They look diverse when you don't understand them.

Same about languages. When you go back again about 50 or 16 years, it was commonly believed by major professional linguists that languages differ so enormously, that each one has to be studied on and so on without any preconceptions. By now we know that that's totally false. It's very much like biology. If you get into the deeper parts of them, it turns out they're very much cast to the same mold. In fact, if that were not true, no child could ever learn the language.

It's got to be, what the child knows, let's say, two or three years old, is way beyond any evidence that's been presented. Right. One example that I gave, for example, you can't learn that. And it turns out that these principles are apparently uniform across all languages. I mean, in a sense, we know that they have to be. Otherwise, language couldn't be learned. So the task of the field is to show what we know must be true.

That is to find the mold to which all of these things are cast. It's got to be there. We have to find it. And the more you look, the more you find it. It's even through the meanings of words. The meaning of a word is so rich that you could never pick it up from exposure. And in fact, we know the children learn words from almost no exposures. at the peak period of language learning, two or three years old.

The kid is picking up a word virtually every waking hour, which means almost one exposure. And they know the rich and complex meaning. Now, the things that you were talking about, like irregardless, they're on the order of table manners. It's like, how should we behave, you know? And you talk differently when you're giving a formal lecture, than when you're talking about somebody on the street.

So we learn those conventions about how you're supposed to talk in different places, but they have almost nothing to do with language. The point is to understand the word regardless requires enormous internal knowledge. You don't have anywhere near enough evidence in your entire life to acquire by induction, anything like the meaning of the word regardless. Or in fact, anything like the word river or creed, or almost

any word you look at, as soon as you begin to study deeply what their meanings are, it turns out it's way beyond anything you get from the environment. And they're the same in all languages. So it kind of has to be the case of the way you couldn't learn anything.

So again, it's kind of like if I look at In x-ray, I see just a lot of noise. If a radiologist looks at it, they see a tumor in a certain place. Well, the infant is like the radiologist. They're genetically primed to look for particular things. So they miss the noise and go after the particular things. And that's true of word meaning, it's true of every aspect of language.

So now we're going to switch gears a little bit and talk about consciousness and maybe segue if you have time into discussion of artificial intelligence and language, natural language processing, and I have some questions related to those. And hopefully we may be able to tie them together in the cognitive scientific miasma that I'm want to construct. But in your 2017 talk at the Science of Consciousness held here in San Diego, sponsored by the University of Arizona, where you're currently located, you have some very interesting perspectives in which at various times I felt hopeful and at various times I felt hopeless. And I'll say there's this running debate, this so-called hard problem of consciousness and the easy problem of consciousness.

And there are those that believe that there's consciousness in every, and subatomic particles perhaps, depending on what definition one uses for consciousness. But I want to start with the very beginning when you really tied into something very important to me, which is the scientific method. And of course, you're extremely well known for using the scientific method, you know, first in the cognitive revolution to use and study cognition in its own right for the first time. In that talk, you coined a term called the Galilean challenge. Do you remember, can you explain what the relevancy of that, topic is. What was Galileo talking about as this fundamental challenge that came from language and

perhaps superseded the challenges that he had employing the physical, the scientific method, perhaps? First of all, on scientific method, I'm sure in your physics department, there isn't a course on scientific. No. But there is no scientific method. It's just being intelligent, you know. I've never once sat down and said, I'm going to form a hypothesis, I'm going to assemble an apparatus. We do it in our lab classes, but you're absolutely right. Yeah, there is not. And some believe there is no scientific method, period.

Not only that in practice, do we not use it, but. The scientific method is making smart conjectures and seeing if they work. I wouldn't interpret it in some scientific method in a nutshell. But the Galileo, remember, this is the beginning. of the scientific revolution, 17th century. And it was a real revolution. If you go back to neoscolastic physics, now they had an answer to everything.

So, you know, you're holding a glass of water with hot boiling water, your hand over it, let it go, the glass falls, the steam rises. We have an answer. They're going to their natural place. You pick a big lead bowl and a small lead ball, and you drop them.

The big one is going to go faster, because that's our experience. You perceive a triangle, the form of the triangle, goes through the air and implants itself in your brain. Well, the scientific revolution began when people decided to be puzzled about those things. They said, why should I believe any of this? In fact, as soon as you think about it, some of them are wrong, like the rate of fall,

Galileo, disproved it by thought experiments, never carried out any experiments, but it was able to show you obvious it's false. One of the things they looked at, and the same with the rest, that's when the scientific revolution began. Well, one of the things they looked at was language. They were puzzled by that, and what Galileo and others were puzzled by. In fact, regarded as kind of an amazing, incomprehensible fact, is that with a few symbols, we somehow are able to construct infinitely many thoughts in our mind and even find a way to get others who have no access to our minds to comprehend what worth of the inner workings of our mind.

that's a miracle you know they're right it's a very hard problem you don't know how to solve it but that's the Galilean challenge how can that take place right

it tends up to the present and now we have parts of it that we can understand other parts remain mysterious so you speak in these talks on consciousness about sort of this

internal system and external system, or the system of making something external. Can you explain, what do you exactly mean by internal system? It sounded slightly ill-defined in that it's very difficult to say, well, here's my internal system and a mechanistic reductionist point of view. But what do you mean by the internal system and the interaction that you make it very precise? I mean, we're just talking loosely. But if we start with the simplest combinatorial operation, as I said, binary set format, we ask how it applies, we take a look at some other conditions. So, for example, there's good reasons to believe that the way the brain works, it keeps to principles of computational efficiency. You have some understanding of those. You bring those in. It tries to limit user resources.

So, for example, one of the striking things about the brain is that it's extremely slow. If you look at the visual system, say the retina, a single cell of the retina is picking up a photon of life. Yes. It's passing a huge amount of information into the system, but the brain is much too slow to deal with it. So it throws almost all of it out some way of keeping the resources limited. to try to work out these notions of resource limitation, computational efficiency, other things you pre-sum begin to get sharper ideas about how the internal system is working. And you can make it quite precise. Then comes the question, how is it coded in the brain?

That's the next question. Notice that's a very hard question for ethical reasons, not for scientific reasons. Remember that the language system is unique to humans. We can't study other organisms. They don't have it. So the kinds of invasive experimentation that have been used to lead to understanding of the neurology of the visual system can't be carried out. We can't carry out experiments with, say, children in isolation and see what would happen, ethical reasons. So you have to, in order to study something that's unique to humans,

almost all the modes of direct experimentation are excluded. They're just not allowed to do it for ethical reasons. Yeah, and that was sort of related to the thought experiment of communicating with an alien that you would avoid presumably the ethical implications, although I'm sure there were awful experiments done during the Nazi regimes on living subjects, maybe not in the era where we could appreciate their impact on consciousness or whatever. But yeah, you're right.

That is a difficult, there is no true way to provide a null hypothesis with which to compare. Is there in physical? I would be clear. You can do neurolinguistics. They have to be smart about it. The experiment I mentioned at the beginning

about the impossible and possible languages and their brain correlates. That's our linguistics. Okay. So you can figure out indirect ways to learn things, but you can't do the experiments that immediately come with mind, like sticking an electrode in Broca's area and see what happens. You just can't do that.

But you can do it indirectly. It's a little like cosmology. You can't go back a couple billion years and say, I'd like to see what's happening. Not yet, right. And it's very much like that, but you can learn things. So going back to your question, at the level of the computational system on how it works, you can get fairly precise. When you ask, how is it coded in the brain somehow, you're running into a very hard problem.

problems, which are limited because you can't do the experiments that come to mind. You have to do indirect experiments. So there it becomes harder. But these are all within the bounds of scientific inquiry. In science, again, turning... Studies and play. Come together on a Windows 11 PC. And for a limited time, college students get the best of both worlds.

Get the Unreal College deal, everything you need, to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft 365 premium and a year of Xbox GamePass Ultimate with a custom color Xbox wireless controller. Learn more at Windows.com slash student offer. While supplies last, ends June 30th, terms at AKA.m.m.m.m.com slash college PC. Back to fine men, but even back to Format and others, there's a notion of what's called the principle of least action, which is an expression of parsimony in nature that the shortest paths, geodesic paths, are taken in physical systems that minimize a certain quantity called the action, which in turn is related to certain dynamical variables that characterize a system in physics. Is there an analog? And that's one of, by the way, the most cherished, sacred principles of all of physics.

In fact, it holds for everything, including the propagation of light and quantum electrodynamics and quantum. field theory, even from the 1600s up until the modern day. Is there an analog in linguistics? I mean, you mentioned that we are forming thoughts, we have meta thoughts, we're throwing out a lot of data. How does the mind know how to do that? Is there an analog to this principle of least action? Yeah, computational efficiency.

Principles of computational efficiency are analogs to the laws of least action. And they show up very immediately in our, I mean, let's take a sentence like, the boys expect to meet each other at the beach. Each other goes back to the boys. It picks the closest thing. Suppose you say, which girls do the boys expect to meet each other at the beach? Each other doesn't go back to the boys. It goes back to something that's not there.

What's in your mind is which girls do the boys expect the girls to meet each other? That's in your mind. And each other goes back to the unheard, the girls. But why don't you pronounce the girls? Principle of least effort. The printer wants to do as little as possible. So it eliminates a lot of stuff. It just does the minimal it can get away with.

It has to pronounce something, or you don't know the question was even asked. So it pronounces just the most prominent thing, none of the others. That leads to major problems in communication. In fact, for people who do automatic parsing, one of the biggest problems is what are called filler gap problems. You hear a word like, rich girls, you've got to find the place where it's not there. and that's a big for this sentence it's easy when you get to more complicated sentence it can be a huge problem so because of computational efficiency the analog to

a law of east action you're getting huge computation communication problems but the internal system is working with maximal efficiency it doesn't erase anything that would be an extra operation that's done for so and in fact this is related to the questions of what we call talking to ourselves, internal language. We're not talking to ourselves in our language. When you think the sentence, which girls did the boys expect to meet each other, you're thinking it the way it's pronounced. You're not thinking what's going on in your mind.

That's inaccessible to you. That can only be understood the way you can understand how you have visual, system works by external investigation. So almost all of our thinking is inaccessible. We're only getting a periphery of it, what's around the printer level. And what's really going on, you have to study as if it's some physical system you have no access to, because there's no way to introspecting to it. If you get introspect, linguistics would be really easy. You just think what it is, but you can't do it because it's all. all inaccessible. Now this bears on the consciousness issue because what we're conscious of

is little bits and fragments that kind of come out from whatever's going on inside. In fact, if you really introspect and you think what's coming to your mind, it's not sentences. It's bits and pieces of this and that and the other thing. But you can make decisions very quickly. microseconds that are complex decisions about a variety of different things. Like you walk into a room, see some guy sitting over there who you wanted to say something to, so you go over to them. But you notice somebody else is sitting there who will be insulted if you say it, so you decide not to.

And then you decide to say something else and so on. And this happens instantaneously. But bits and pieces of the conversation that you're having. doing do reach consciousness. So you'll get a fragment of this, fragment of that, and so on. What what reaches consciousness is a very superficial, partial reflection of the internal computation that's going on. That means if you want to seriously study consciousness,

you're going to have to learn about the internal processes that are putting forth the bits and pieces that pass through that. filter and reach consciousness. Very small, superficial amount. Here's some language from a mammal in the background there. Working from home. Yeah, I don't know if there's a way to,

I don't want to muzzle the dog, but if a dog can keep it down while we're on the podcast, that would be great. So I want to turn now from, well, consciousness in the state of something you said at the science of consciousness. Actually, you said this, the next year in 2018. And I wonder, you know, if there was a change between 2017 and 2018, perhaps not.

Perhaps I missed it. But you said the following, the inner workings of the mind are inaccessible to consciousness. That's a very profound statement. What do you mean exactly by that? And since I don't remember it in 2017, you said in 2018, maybe there's a chance you no longer, you no longer, you know, longer maintain that statement, but you said the inner workings of the mind are inaccessible to consciousness. What does that mean? I just gave an example. The inner workings of your mind for the sentence that I gave, produce the sentence, which girls, the boys expected the girls,

to meet each other. That's inaccessible. We can find it indirectly by studying the way words like each other work. They do work by picking out the nearest element that's, again, least effort. But they're doing it on something that you can't be conscious of. You can't be conscious of any of the operations that are taking place. So the inner work is, it's very much like the inner workings of your,

like you have a gut brain. Yeah. It's what's called the enteric nervous system. huge nervous system that's carrying out these huge operations for keeping your body functioning. You can't introspect into it. The only thing you know about it is I have a stomach ache. Nothing's wrong. And it's very much the same with the brain that's in our head.

We can see little bits and pieces at the surface, but we can't figure out what we can't interest. We were totally unconscious of what's going on. And there's no way to become. So the same is true, the meanings of the simplest words. I think the first case that was studied in the history of science was Heraclitus, the pre-Socratic. He asked a very profound question, how can you cross the same river twice?

Okay. You think about it, it's not a trivial question. The second time you cross it, it's totally different. It's a different river, right? The same river. You start playing with this. You realize that you could make radical changes in the river. It would still be the same river.

You can make tiny changes, like a phase change that switches it to the glassy state and then run cars on it. It's not a river. It's a highway. Almost indetectable change, but it's not a river. Huge changes. It will still be a river. Now, every infant knows this, and it's very complex when you look into it.

You can't introspect into it. You have to do experiments to figure it out, like these thoughts are thought experiments. And it's with every word in the language, and of all the constructions in the language, all the methods for constructing the thoughts that we're producing constantly, totally beyond the level of consciousness. So it may sound strange, but if you think about it. for a minute. It's kind of almost obvious. So we spoke about, you know, the ethical implications of testing, you know,

consciousness and impact and human cognition. I want to turn to artificial intelligence now and ask you, first of all, are there applications that were artificial intelligence could shed light or perhaps already has shed light on these problems of consciousness, in your opinion. Artificial intelligence divided into two fields. One of them, which Marvinsky was interested in, was trying to find out something about the nature of intelligence. That's science. It's indistinguishable from cognitive science. It just happens to be using different devices. So it's doing it by modeling with computers instead of modeling on paper. But it's basically some cognitive science.

That's one part of artificial intelligence. The other part is engineering, trying to construct something that's useful, like, say, a Google translator. It's done by brute force, absolutely brute force. No scientific interest whatsoever, kind of low-level engineering. Machine learning, right? Most of what's done in deep learning is brute force. You try to do massive computation of rapid computation of huge amounts of data and see if a pattern

emerges. Okay. It's okay. Like a Google translator is useful. I'm glad to have it. I'm glad to have a bulldozer. But it's not, there's nothing good with science. Now, if you ask about the language side, it's almost impossible to answer.

because the part of AI, the sort of Minsky thought that is essentially indistinguishable from cognitive science. So the answer is automatic. Language is just part of it. Now, the other side doesn't have anything to do with science. So it basically doesn't tell you anything. Like if you have a word process, it would be like not a mind. Now, there's a bit part of it, which is a pain in the neck. It constantly tries to predict the next word you're going to say.

It's in nuisance because it gets no way of typing. That's an AI, a deep learning program. You know, you study massive amounts of text with some supercomputer, do a lot of statistical work. You can get a pretty good prediction of what the next word is often going to be. It's totally meaningless. It's like looking at billions of chemical experiments. And getting to the point where you can predict that if you mix these two things, it's going to turn blue.

Okay. Don't tell you anything about chemistry. That's just nonsense. Right. So that's statistics. One scientifics of AI, the other part is just science. When we think about things, well, first of all, when I want to get back to the predictive text, one of my friends is a very popular podcaster named James Altucher. he says that, you know, one day he was doing that and he was frustrated by it as it sounds like you are.

But then he realized actually this AI was helping him be a better natural human being intelligence. In other words, it was telling him things like suggesting, how are you? You know, how is your day? And thank you so much. And these are things that he wouldn't say. Why are you bothering me? But it was actually, you know, ameliating or moderating some of his more. gruff personality traits. And I wonder, yes, it's a nuisance, but then, you know, there are,

there are, you know, artificial intelligence has a role in, in some sense, to do, to make predictions based on experience, and that experience can only come through, you know, the brute force approach, at least for now. But I wonder, you know, when we speak about artificial general intelligence and so forth, there's a famous Turing test. And I wonder, you know, nowadays you've probably seen there are these captures, you know, there are these images where your computer asks you to prove that you're a human being. It's sort of, you know, an inverse Turing test, like you have to prove you're a human being to a computer, which is sort of a little bit of an inversion of the classical Turing test. But does, you know, does language

play a crucial role in the Turing test? Like I can't imagine my two-year-old, you know, being able to tell the difference between a D.D. recent AI that one of my undergraduates could program versus the most sophisticated deep mind that Google might have currently. So it seems to me that the ability to pass the Turing test almost is dependent on the cognition or language abilities of the human operant at the terminal. What do you feel about the Turing test as this modality to distinguish general artificial intelligences?

Well, let's begin by asking what Turing thought about it. So if you look back at his famous 1950 paper on machine thinking, he says, the question whether machines can think is too meaningless to deserve discussion. Okay, that's what Turing thought about it. He thought that the imitation game, as he called it, could be a useful device for stimulating better construction of machines, of software, and so on. And I said it might in 50 years, you'd guess, modify the way we think about thinking. But it's a question where the machines can think is too meaningless to deserve discussions. It's kind of like asking whether submarines can swim. You want to call that swimming, okay, they can swim.

In fact, languages differ in these. In some languages, airplanes fly and other languages they glide. These are just uninteresting conventions. Now, it takes a Turing test itself and go back to the 17th century, the origins of modern science. They had something like the Turing test. Descartes asked the question. He was part of this Galilean challenge. He was Galileo and many others.

Descartes asked, how can a person, carry out the normal creative use of language. You know what to say in particular circumstances. You're producing sentences constantly, which are novel. You never heard them before, nobody ever heard them before, but others can understand them. They're appropriate to situations, but they're not caused by situations.

You could have said something else. So, as they put it, you're inclined to say certain things, but not impelled. This property of being able to, and then some of his followers, Jacques de Courtaire, another Cartesian, proposed test, experimental tests, said, suppose there's another creature that looks like us. We want to find out if he has a mind like ours. So we run experiments to ask him, would he say the kind of thing that's appropriate in particular circumstances? That's the Turing test. But it was different in the 17th century.

There it was science. For them, remember, it's a question of existence. There's a mind, which is a thing in the world. There's a body, which is a thing in the world. and we want to know whether another creature has the mind. It's like asking, does he have a liver? It's asking a question of the physical sciences. That's right.

For the Cartesian, for the Galalans, the analog of the Turing test, it's a straight scientific question. For Turing, it's not a scientific question. It's a way of stimulating your imagination or something like that. Thought express. So in a way, the 17th century tests were much more, serious. But this, you know, going back to your computer that tells you something, yeah, that's fine.

I mean, if Alexa helps you think of something, who cares. But there's no science involved. Right. It's like saying my electric stove works. Is that parlay with or dovet with your well-known views on the Elon Musk's Neurlink project, where you've said that, you know, trying to move your arm, you know, with a neuro-embedded chip like Neurrelink is perhaps feasible at some point. But to find out what you're thinking, there seems to be, you claimed in 2018, I believe, that there's no way to do that because we don't understand how to proceed. And I think that, I don't think your views have changed much, right? Only that, we don't even know if we're looking at the right thing.

How so? Thinking may not involve neural nets. In fact, they're a pretty good reason to believe that it doesn't. The neural nets for one thing are neural transmission is pretty slow as we were, you know, by the relevant criteria, by the criteria of what we were told about before, how rapidly you think by that criterion, as known back to Helmolts, in fact,

the neural transmission is pretty slow. Furthermore, neural nets. don't have the right architecture. You have to, what we need is something like touring architecture, something that has basically the control unit of a touring machine, right dress, so on. You can't do that in the romance. They just don't have the right properties.

That's why Stuart Hammeroff, what you mentioned before, is looking at things like micro-touch Things like microtubules, things on the internal structure of a neuron, which has vastly more computing power. Roger Prenrose is working on this. The main work on it was done by Randy Gallastom, very good neuroscientists, has done very interesting work arguing what I just said, I'm just quoting him, that the neural net systems are just the wrong place to look.

They don't have the kind of architecture, which is involved in thinking. We have to find something else. It might turn out to be at the molecular level, at the level of RNA, molecular level you're really getting massive possibilities of computation. So maybe just duplicating a neural net will tell you nothing because you're not looking in the right place.

We don't know. I mean, the thing to do is do the science first, then worry about the engineering. So I wonder now if we can turn to the topic of the university and academia. And I always like to ask guests such as yourself who are public intellectuals plus academicians for many, many decades. I want to ask you what you see is the future of the university, and especially in this era of COVID and so forth. And then after that, a follow-up question will be if there was a Chomsky University, what would you have on offer there? I would treat it the way an old friend of mine, a physicist who you know very well, at MIT treated it, Vicki Weisskopf.

He was famous at his freshman introductory courses when a kid would ask him, what are we going to cover? the semester, he would say it doesn't matter what we cover, it matters what you discover. That's what an educational system should be. And I think you can extrapolate from that in every direction. So the worst kind of education imaginable is what's called teaching to test what we do in the schools. Every one of us knows you've had a boring course where you'd bother to learn the stuff. and you know, aced the best. And a week later, you forgot what the course was about.

Okay. That's what we impose on children. The worst possible kind of education. The right kind is what Weiss Koff was talking about. So the right kind of education, let's stay in a science course. And there are a very good program. There's one example. Take a kindergarten.

Give the kids, each kid in the kindergarten is given a shell. And on the shell there are several things. A bead, a piece of grass, a seed, a bunch of things. And then the teacher poses a problem. Which one of these is going to grow? And then the kids have a scientific conference. They try to figure out some way to decide which ones can get would grow. There's a little supervision from the teacher, you know, sort of keep it in the right direction.

But finally they figure out, you know, one way to do it is put it in the earth and water it. And they finally they've figured out how to do it and something grows. At the end, the teacher gives them each a microscope. It splits the seed in half. You can see what's inside it. It's making it grow. That's education. Teaching to test.

You could say, here's the answers, learn them, repeat them in a test. Zero effect. You don't, you don't learn anything, and you don't understand how to learn, which is the most important thing. I should say all of these topics were discussed in the 17th and 18th century, and they used the model of pouring water into a vessel, one kind of teaching, the kind they said is absolutely no good. Right. I want to just pour water into a vessel. That's used an geeky vessel.

The right kind was, as you described by Vilho von Humbold, founder of the modern research universe. What it is is the teacher lays out a string along which the student follows in his or her own way. Some structure and guidance. But the best math courses I ever took in my life. graduate course and real variables. He would come into the class and clean the blackboard and write something down on the blackboard

and say, is that a theorem? And the rest of the class would be trying to figure out if that's a theorem based on our reading for, you know, so can you get a lemma from which it would be proven, maybe so, can you figure out a way to do that? And that's education. And you can do it at every level from kindergarten, humanity is sending. Ambition comes in all shapes and sizes.

At First Citizens Bank, we roll with your goals because we're built for what you're building. Fit for your ambition for Citizens Bank. Yeah, I agree. As you said, the analogy of pouring water into a vessel, literally the word educate comes from Latin Educare, which means to draw out, as I remember, which is kind of in concert with exactly the way you're describing it. My friend Mario Olivier, who's an astrophysicist at Johns Hopkins and Space Telescope Research Institute, he has a book called, you know, why, what makes us curious. And he claims, you know,

epistemic curiosity is the key to education. And in fact, you know, that instead of teaching your kid, you know, why the inverse square law of gravity, he says, no, don't start with that. Start with dinosaurs. You know, ask them, do you know how the dinosaurs died? And every kid loves dinosaurs, and they're going to want to know, and especially to tell them an asteroid, you know, is likely the reason for their demise. And then you're just increasing the level of tension in the spring, and it just wants to unwind so badly and release that that is the way to educate. And yet we're stuck in this model. I mean, the modern university system is, you know, know, far better than I am. It goes back thousand years to Bologna and perhaps before in the,

in the, you know, end of the beginning of the second millennium. And I wonder, you know, that hasn't changed. And I wonder with things like AI and you're obviously not in a super, as your colleague, former colleague at MIT, Max Tagmark is, as perhaps as sanglin about the benefits or the future potential of AI. And yet, you know, why should you know, why should. somebody take a class with Brian Keating in physics at UC San Diego, when he or she someday could take a class with Galileo himself or Marie Curie or whoever? Is there a needed change in the university system to break the sage on a stage, you know, scratching one rock of chalk on another blackboard rock? Is this model due for a change? Or do you feel like the in-person learning model that we've had for

a thousand years will persist. Well, first of all, I think in the best cases, it does persist, like the cases I mentioned. The mass class, I took, Vice Copts, physics class, Peter Gordon case, and many others. I was at MIT

most of my life. I'm now at the University of Arizona, and there's an interesting change. MIT is very old-fashioned. I was able to use blackboards and chalk. Now it's not chalk. It's some other gimmick, but you could write things

on a blackboard. I've noticed I'm talking at other universities in here. You've got to use PowerPoint. Yes. I can sort of learn how to use it. I find it much easier just to go to the blackboard

and think. Maybe what I thought about in the PowerPoint is what I feel talking about. Also, interact. So if the student comes up with something, you can write it in the black word, let's talk about that. I think, I'm not entranced by these educational advance, I'm not sure. Young people seem to like them. It's kind of key to the video culture that they live in.

I even see students like it if the professor has a PowerPoint and reads off the thing on the screen. I don't understand that. I like it the way my old math teacher did it. But I don't think that's really the issue with the universities. I mean, you can do a good class anyway. But it's the question whether, like the person you just mentioned, are you going to encourage curiosity or dull it?

I mean, the children are naturally curious. They're always asking why does it work this way? What's the answer and so on? Now, you can either stimulate that curiosity or kill it. And unfortunately, a lot of the educational system kills it. But it doesn't have to. It can stimulate it in all the ways we've been talking about. Yeah.

And that's, if you're a country schoolhouse, a little red schoolhouse with just a blackboard and chalk, and every grade together, you can do it there. You can do it in a fancy university or classroom with all kind of extra bells and whistles that you can push. Here's a random question for you. Have you ever meditated? And I'll explain why in a second. Have you ever meditated? They say if you don't have time for it, then you need to do three hours a day.

So the reason I ask is that one, and I've tried this, and it's become increasingly popular in many different realms from even the military to, you know, to, you know, peak performers. It's not just for Deepak Chopra who helped us meditate when we were at the science of consciousness together. And part of the goal is to stop your inner monologue. And I wonder, you know, there are those that claim when you do such a thing, you achieve enlightenment, and I'm not such a huge fan of that. But the mass industry of the meditation industry, literally millions, billions of dollars, perhaps apps and books and seminars and gurus and mantras, I've tried to dial into what it really means. And it seems like you're trying to stop the inner monologue. You're trying to stop this incessant fire hose. And it's almost as if the human being is, some are ashamed of it, or it's negative in a way that, you know, saying we're much more comfortable talking about, you know, sex or, you know, things that are used to be taboo, maybe. But people don't go around saying every thought that they have. And I wonder, why is that? Why is the inner monologue sort of the last taboo that, you know, these, these incessant thoughts that are bombarding every human brain? Why is that so taboo to speak about? Or is it? Maybe it's not.

I work on the inner mind all the time. Students of language, that's what they're doing. Studying visual perception, that's what you're doing. Anything in the cognitive sciences, you're studying the inner mind. What you're talking about, you know, the bits are, right of all, we're not aware of their inner minds. We're only aware of bits and fragments that come out every once in a while. There's a machine there, and ones and that spits out a little bit

this and that. That's what we're aware of, but it's not the inner mind. The inner minds, you can only study from the outside, as it's the same way you study the gut brain. Can't, no introspection. But sure, we can study it. On the other hand, why don't you go around talking constantly whatever's on your mind? Well, there are people who do, they're called children. If it wasn't your three-year-old kid, you're talking constantly, anything that's on their mind that's the same. They haven't learned and keep it quiet yet. So it's okay with a two-year-old.

It would be a pretty awful world of the 40-year-olds. I keep my inner child and permanent timeout. So we're almost done here. I appreciate your time so much. I want to just finish up with a question from a friend of mine, an intellectual as well, mathematician Eric Weinstein. And it kind of relates to, you know, this perception of you as this controversial figure.

And just recently there was a letter that you were a co-signatory of. I believe the title of it is a letter on justice and open debate. And you and J.K. Rowling were the two, you know, kind of featured co-signatories, although there were hundreds of people, over 100 people. And I wonder what, and Eric, points then asks the following question, will liberalism survive this diversity movement or the moment that you guys were sort of decrying in this letter that you wrote, the open letter on justice and debate? First of all, let me mention, turn to the point that we were the two mentioned. That's a sign of the utter irrationality, incurable irrationality of the intellectual culture.

anybody who thinks for one second can see that you don't evaluate a statement by the signers. If you did, there would never be a statement for a very simple, logical reason. I'm sure you get plenty of statements you're asked to sign. You don't know who's going to sign them later. So if you care about who's going to sign a letter, you'd never sign it. So therefore, there wouldn't be any statements at all. So even to pay attention to the signments reveals profound irrationality.

What matters is the content of the statement makes no difference who signs it. For elementary reasons of elementary logic, just what I just said. So I don't care who signed it. You can't know who's going to sign it's impossible. The fact that this analysis, The Gennadine statement received a flood of reaction is very interesting.

It's a simple, straightforward statement, almost two elementary to sign. It says what everybody ought to believe. That doesn't mean it's not important to say. There are tendencies in the university. You're all familiar with them, which are limiting discussion. We can say that's not a good thing. thing, period. There shouldn't be one article in any newspaper referring to it. The fact that there's, so the only interesting thing about the statement is the reactions.

Why are there reactions to such an elementary comment? And why do they focus on signers when if anybody thinks for a moment, if they can figure out that if you pay attention to signers, there'd never be a statement on anything. Okay, so I think we're looking at an interesting case of the radical irrationality of the intellectual culture. It's about the only interesting thing to say about this. So is there hope? Do you feel that this irrationality is going to, is on the upswing, so to speak, from your perspective, or is it likely to dissipate? And what role, if any, does linguistics or have to say about it? I'll mention why in a second I said that, but do you find it's diminishing or do you feel fear for the worst that it's going to get more and more ideologically entrenched that even these anodyne statements in your words cannot be countenanced?

Well, I think there's an interesting thing going on. You go back to earlier years, there was a very high degree of uniformity. So, for example, I could give you examples from my own experience. Anything that shifted a little bit from the ideological mainstream was just canceled. Can't get in. Now there's been a good thing that's happened is there's more concern for other issues that weren't talked about before. Women's rights, rights of blacks, you know, human rights. and rights generally, a lot more

the diversity. Now, one of the effects of diversity is it can be overdone. So you're getting a kind of a confluence and it's got good things, it's got bad things.

We should be rational enough to pick out and emphasize and develop the good things and to put to the side the bad things. And the question is, of us can we do that. Do you feel that there is a role that perhaps linguistics might be uniquely capable of,

or language in general might be capable of apprising in the sense that, you know, it's sort of a trope or whatever, that when you hear an accent from some place, you know, the South, it's, you know, there's Jeff Foxworthy makes a bunch of jokes about this. He's from the South. He'll say, you know, the last thing you want is your brain surgeon to say, yeah, what you're going to do is go down and cut open, you know, and they may be fully qualified, and, you know, he's, he's joking about that. But there is, you know, in some of that, there's a stereotype that certain accents sound uneducated, British English, for example, sounds sophisticated,

even if the person who's uttering the words might be a total ignoramus. Why does, why do we have these reactions? I mean, why is that encoded? Why do we encode a prejudice based on the sound, It's a meta form of the language itself, not even the structure of grammar. It's much more than language. If you go to a formal party dressed the way you are now, it would be improper. Yeah. Not because it's wrong, just because that's the convention. So there are certain hierarchies of power and authority, which say you've got to behave like me,

so you don't talk with a Southern Cracker accent and a, you know, in a formal occasion. It's not, we shouldn't accept, it's not a problem of language. It's a problem of the authority structures. They shouldn't have that authority. You say it's on your business how I talk, you know, I talk the way I talk by. And it's the authority structure that's the issue.

So let's go after that, not the superficial symptom. Like what clothes you wear, or did you comb your hair the right way, did you remember shave this morning or whatever? That's no interest. What are you saying? That's what matters. Kind of like this statement. What's in it?

Not who decided to sign it. We have to just overcome these prejudices. They're like a lot of others, you know. A lot of them are quite pernicious. A lot of them we have. overcome. Things that were considered quite normal, not many years ago, are considered totally unacceptable now. Plenty of them. And that's good, but we can't overdo it. You can't get to the point where nothing to be said without tiptoeing on eggshells. You've got to find the right

boundary between those. And that holds not only for language, but for all kinds of behavior. I don't think linguistics really has anything to say other than every language is the same as every other language. Okay, that's true. Like if all the power were in the hands of those southern crackers, we'd have to talk like that. So, Noam, there's two more questions, if you'll beg my indulgence here. The first question that I like to ask all of my guests on the Into the Impossible podcast relates to Sir Arthur C. Clark's book, 2001, a space odyssey, made into a film by Stanley Kubrick. You'll probably remember in the film, there were these monoliths, there were these very imposing objects that were found on the

African savannah by some primates, and then later found on the moon's surface. And they're allegedly left in the book series. They're left as a sort of a way to communicate messages to humanity placed by an ancient civilization that was obviously far advanced of where we now, but meant to be discovered at a certain time when humans were capable, say, of going to the moon, for example. And so my question for you, the first of these two questions relates to a time capsule, if you were able to make a time capsule that was going to last a billion years like this monolith, what would you put on it or in it or what synoptic view would you like to engrave and code and crypt into such an object to last for a billion years?

Well, actually, that problem is very real, since it's very likely that humans will extinguish themselves within a couple of generations, the problem is very much alive. Either nuclear war or environmental catastrophe, if we continue on our present course, we're not going to survive. So the question is not abstract. The first thing we should do is try to see if we can avert those outcomes. It's still time to do it. That's the major question in human history.

Okay, I suppose we can't. What you put in them is the greatest works of science, of art, of literature, or any aspect of human achievement. That's what new people have been in years now for now. should be striving for on some planet, maybe not this one. That's right. And the last question, Nome involves going backwards in time, not forwards in time. And as I mentioned, the name of this podcast is called the Into the Impossible podcast,

in allusion to Sir Arthur C. Clark's third law, so-called third law, which states, the only way of discovering the limits of the possible is to venture a little way past them into the impossible. and accordingly I would like to know in your life what aspect of your life perhaps you as a 20 year old a 30 year old as a young academic what perplexed you seemed impossible and then makes sense to you now with the retrospect of time so what sorts of advice would you give to your former self perhaps as a 20 or 30 year old I would give the advice that I In fact, gave to myself.

At the time, the field domain didn't exist. In fact, the first book I published, submitted the MIT Press and got a very sensible reviewer reaction saying, this field doesn't exist, can't publish it, was right. But the advice, I didn't bother giving advice. I just said, I don't care. I'm going to do what looks interesting. And that's the right advice.

If it doesn't work, to bed does work. Okay. What looked impossible at the time was what we were talking about. At the time, it looked as if languages just differed totally from one another. Each one had to be looked at it alone in its own way. When you think about it, that can't be true. Plus, if it were true, nobody could ever learn language. since what they know is way beyond any evidence.

So it's a real paradox, and it didn't seem to be any way to solve it. In fact, I think by now we're just about getting to the point where maybe we can find an answer to it. All right. Well, I hope to have you back on the podcast when that happy day comes. But for now, Noam, I want to thank you for your time. I want to wish you a happy and maybe cool summer in Arizona. I don't know if that's it. That's impossible, perhaps, to envision.

I'm sure we got below 100. Very good. Thank you so much for joining us on the Into the Impossible podcast. Yamava Resort and Casino at San Manuel is California's number one entertainment destination for today's superstars. Catch the Jonas Brothers return to the Yamava Theater stage on April 30th, the powerful vocals of Demi Lovato on May 17th,

and the signature Southern Country Rock of Eric Church on July 19th. Tickets on sale now at yamava Theater.com, only at Yamava Resort and Casino, celebrating its 40th anniversary. You in? Must be 21 to enter.