With this new Jeffrey Epstein Spectacle drop, everyone is once again buzzing about Noam Chomsky and his relationship with The Pimp to Our Unfuckable Elites.
I was going to comment on this, but then saw that you addressed it in your closing paragraph.
Still, a small observation, just because I like discussing this:
"It’s funny, too, because Chomsky’s detractors were proven largely right: his search for Universal Grammar has been discredited. The way LLMs handle language is through pure statistics and brute forcing massive amounts of data. So the paranoid Pentagon bureaucrats had backed the wrong horse with Chomsky. But it didn’t matter. Chomsky was still very useful. He helped push the larger field of psychology in the right direction: towards studying minds as computers."
Chomsky's Universal Grammar was _not_ discredited by LLMs working differently. It was, rather, discredited because there is no experimental evidence at all for the existence of Chomsky's "language module" in the brain, and because his theory's predictions regarding the behaviour of the world's languages is not what is observed. _And_ because people working in current neuroscience and biology don't believe that the brain behaves _at all_ like a computer, and certainly not like a digital one (as Chris Knight also points out).
LLMs couldn't discredit Chomsky's theories, because they also don't work at all like the human brain - given that they do inference from humongous amounts of data, while humans learn languages from rather limited amounts of input. Also, LLMs don't have anything at all of what we usually call "insight" or "understanding".
And, of course, studying minds as computers was never the "right" direction for psychology for anybody except the Pentagon people and technocrats, but that's also what you write at the end. Treating the mind as a computer is convenient if you want to simulate it on one, but this metaphor gives no real understanding of what's actually going on in people's messy, non-digital, relational minds.
The parallel to Silicon Valley's much-lauded concept of "disembodied mind" is striking, though: Silicon Valley stars from Stewart Brand to John Perry Barlow used to rant on about the possibility of "disembodied mind", or even "uploading your conscience". But what we understand as "mind" or "consciousness" can probably only ever be embodied.
Which is a nice philosophical discussion for another day :-)
there's a grim irony to how seemingly all these soft power weapons funded by the DoD boomeranged right back into the domestic populace. Machine translation didn't beat the Russians or the Chinese - it has eviscerated the fields of translation and interpreting altogether in my lifetime and within about a decade you're simply not going to have people learning other languages, which has the net effect of further entrenching divisions with other foreign cultures.
Some would argue that this is in fact been the entire point and they are right - since what use is the department of defense (or DoW under Hegseth i suppose) if there is peace.
Bravo, Sasha. I worshipped Chomsky in part because the very messiness you correctly ascribe to us mere mortals also makes us want to believe in immortals (superior beings) of the Chomsky type. We mortals may be confused and constantly contradicting ourselves, but Noam is God-like in his unfailing dedication to reason, logic, science. . . as illustrated in his calm, cool and collected demeanor. We mortals become slaves to our empathy, but Chomsky is like a computer. We rant and rave about injustice; he never rants or raves about anything. It's taken me seven decades to figure out that the best way to explain my political activism o others is not as a journalist (though I am one) or an intellectual (that too) or "high IQ" (yup, that's me) but as an artist who loves gardening and raises chickens.
From what I understand LLMs have no theory of mind or linguistics, other than just ingesting as much text and speech as possible to come up with patterns.
Me too. I think also he modified his theories so much over time to deal with challenges to them that they are not even the same theory anymore. That book Decoding Chomsky is really great. Not an easy read at times. But great.
Agreed, it's great! I disagree with Knight a couple of times, I suspect he has some specific ideas about the evolution of language that are also not supported by experimental data, e.g. from archeology. But all in all it's a great book. I was somewhat stunned to read that Chomsky has really argued that his theory can't be invalidated experimentally because he has thought it out with his "scientific mind" ... that is obviously *not* how science is supposed to work.
[ I am going to give in to my pedantic tendencies, I'm afraid. ]
On the Surv.Valley quote that Chomsky “theorized that language could be boiled down to LOGICAL expressions:”
That sounds more like Montague semantics (a translation of natural language into mathematical logic) than Chomsky's theory of syntax. FORMAL expressions would be a better fit, I think (where “formal” is to be understood as in Formal Languages & Automata, up into the 1980s the mainstay of theoretical computing science, with a foundational impact of Chomsky's generative grammar. This theoretical focus of comp sci was reflected, on the more practical side, in the over-emphasis on syntax and syntax analysis in the design of programming languages resp. compilers [code translators].)
By the way, the “Prussian philosopher” (Wilhelm von) Humboldt mentioned at the end of the excerpt from Understanding Power was a linguistic Kantian apparently (I am going by hearsay) acknowledged by Chomsky as a precursor concerning the creative aspect of language use, although Humboldt's view of language as a dynamic, creative process instead of a fixed set of rules seems to fit generative grammar at best superficially.
I think "formal" is better—maybe "algorithmic," too. I only learned Chomsky's first cut at syntax, the combo of phrase structure rules and transformation rules, and it seemed to me to be all about which patterns fit (phrase structure rules) and then having rules that can describe what can be done with those patterns (transformation rules). Since it's formal and algorithmic, it can be implemented in a computer.
Your post strikes well. I've always felt uncomfortable with Chomsky. He is useful, in garnering facts, but his whole project seems to be an attack on the self-deception of the American intelligentsia. Something that for me is rather boring. Marx or the Old testament I would think does that with a bit more wit and levity.
But recent attacks on him seemed overshot. He has always been, so far as I've been concerned, a figure of the liberal-left. He has been useful and introducing people to somewhat counter-hegemonic thought, but I wouldn't describe him as someone you would point to as a leading light. Norman Finkelstein, whom I admire, of course admires him perhaps excessively. Even John Dolan, whom you and I would think all of us admire, speaks highly of Chomsky on a couple occasions. But exposing over and over again that the US is hypocritical seems redundant. Chat GPT perhaps could have done all his work for us. A collation of lies. This shouldn't be a truth that needs exposing.
This is the most measured take on Chomsky I've come across since the Epstein revelations. I think you're right that his split personality regarding DoD money is a lot more important to viewing what he did (as well as what he didn't do!) than the mentions in the Epstein files, which seem like small beer. He (or his wife!!!) should have known better, and describing the MeToo movement as hysteria was dreadful, saying ". . . [it] has reached the point that even questioning a charge is a crime worse than murder"—What?!?! Horrible, but the reach of his expression of that sentiment is (thankfully!) limited. But by word and deed at the start of his career, he definitely normalized living off DoD money, and that's a problem we're still living with today.
I was going to comment on this, but then saw that you addressed it in your closing paragraph.
Still, a small observation, just because I like discussing this:
"It’s funny, too, because Chomsky’s detractors were proven largely right: his search for Universal Grammar has been discredited. The way LLMs handle language is through pure statistics and brute forcing massive amounts of data. So the paranoid Pentagon bureaucrats had backed the wrong horse with Chomsky. But it didn’t matter. Chomsky was still very useful. He helped push the larger field of psychology in the right direction: towards studying minds as computers."
Chomsky's Universal Grammar was _not_ discredited by LLMs working differently. It was, rather, discredited because there is no experimental evidence at all for the existence of Chomsky's "language module" in the brain, and because his theory's predictions regarding the behaviour of the world's languages is not what is observed. _And_ because people working in current neuroscience and biology don't believe that the brain behaves _at all_ like a computer, and certainly not like a digital one (as Chris Knight also points out).
LLMs couldn't discredit Chomsky's theories, because they also don't work at all like the human brain - given that they do inference from humongous amounts of data, while humans learn languages from rather limited amounts of input. Also, LLMs don't have anything at all of what we usually call "insight" or "understanding".
And, of course, studying minds as computers was never the "right" direction for psychology for anybody except the Pentagon people and technocrats, but that's also what you write at the end. Treating the mind as a computer is convenient if you want to simulate it on one, but this metaphor gives no real understanding of what's actually going on in people's messy, non-digital, relational minds.
The parallel to Silicon Valley's much-lauded concept of "disembodied mind" is striking, though: Silicon Valley stars from Stewart Brand to John Perry Barlow used to rant on about the possibility of "disembodied mind", or even "uploading your conscience". But what we understand as "mind" or "consciousness" can probably only ever be embodied.
Which is a nice philosophical discussion for another day :-)
there's a grim irony to how seemingly all these soft power weapons funded by the DoD boomeranged right back into the domestic populace. Machine translation didn't beat the Russians or the Chinese - it has eviscerated the fields of translation and interpreting altogether in my lifetime and within about a decade you're simply not going to have people learning other languages, which has the net effect of further entrenching divisions with other foreign cultures.
Some would argue that this is in fact been the entire point and they are right - since what use is the department of defense (or DoW under Hegseth i suppose) if there is peace.
I can categorically state that I have no ties to Epstein or Maxwell.
Thank you for your attention to this matter.
Excellent article.
you would say that
Is Kevin Bacon tied to them?
If so, then you are too. ;)
(I swear I didn't know I was ;) .)
Just manufacturing more consent. Great piece, Yasha!
Bravo, Sasha. I worshipped Chomsky in part because the very messiness you correctly ascribe to us mere mortals also makes us want to believe in immortals (superior beings) of the Chomsky type. We mortals may be confused and constantly contradicting ourselves, but Noam is God-like in his unfailing dedication to reason, logic, science. . . as illustrated in his calm, cool and collected demeanor. We mortals become slaves to our empathy, but Chomsky is like a computer. We rant and rave about injustice; he never rants or raves about anything. It's taken me seven decades to figure out that the best way to explain my political activism o others is not as a journalist (though I am one) or an intellectual (that too) or "high IQ" (yup, that's me) but as an artist who loves gardening and raises chickens.
Very interesting, I had thought recently that his linguistic theory aligned with AI
From what I understand LLMs have no theory of mind or linguistics, other than just ingesting as much text and speech as possible to come up with patterns.
Yeah, I guess I meant that our speech operates like it’s been coded but I get confused trying to understand his theories.
Me too. I think also he modified his theories so much over time to deal with challenges to them that they are not even the same theory anymore. That book Decoding Chomsky is really great. Not an easy read at times. But great.
Agreed, it's great! I disagree with Knight a couple of times, I suspect he has some specific ideas about the evolution of language that are also not supported by experimental data, e.g. from archeology. But all in all it's a great book. I was somewhat stunned to read that Chomsky has really argued that his theory can't be invalidated experimentally because he has thought it out with his "scientific mind" ... that is obviously *not* how science is supposed to work.
I’ll have to read that. Thanks for the recommendation.
Recommended reading: THE OBSOLESCENCE OF THE HUMAN, by. Günther Anders.
[ I am going to give in to my pedantic tendencies, I'm afraid. ]
On the Surv.Valley quote that Chomsky “theorized that language could be boiled down to LOGICAL expressions:”
That sounds more like Montague semantics (a translation of natural language into mathematical logic) than Chomsky's theory of syntax. FORMAL expressions would be a better fit, I think (where “formal” is to be understood as in Formal Languages & Automata, up into the 1980s the mainstay of theoretical computing science, with a foundational impact of Chomsky's generative grammar. This theoretical focus of comp sci was reflected, on the more practical side, in the over-emphasis on syntax and syntax analysis in the design of programming languages resp. compilers [code translators].)
By the way, the “Prussian philosopher” (Wilhelm von) Humboldt mentioned at the end of the excerpt from Understanding Power was a linguistic Kantian apparently (I am going by hearsay) acknowledged by Chomsky as a precursor concerning the creative aspect of language use, although Humboldt's view of language as a dynamic, creative process instead of a fixed set of rules seems to fit generative grammar at best superficially.
I think "formal" is better—maybe "algorithmic," too. I only learned Chomsky's first cut at syntax, the combo of phrase structure rules and transformation rules, and it seemed to me to be all about which patterns fit (phrase structure rules) and then having rules that can describe what can be done with those patterns (transformation rules). Since it's formal and algorithmic, it can be implemented in a computer.
Your post strikes well. I've always felt uncomfortable with Chomsky. He is useful, in garnering facts, but his whole project seems to be an attack on the self-deception of the American intelligentsia. Something that for me is rather boring. Marx or the Old testament I would think does that with a bit more wit and levity.
But recent attacks on him seemed overshot. He has always been, so far as I've been concerned, a figure of the liberal-left. He has been useful and introducing people to somewhat counter-hegemonic thought, but I wouldn't describe him as someone you would point to as a leading light. Norman Finkelstein, whom I admire, of course admires him perhaps excessively. Even John Dolan, whom you and I would think all of us admire, speaks highly of Chomsky on a couple occasions. But exposing over and over again that the US is hypocritical seems redundant. Chat GPT perhaps could have done all his work for us. A collation of lies. This shouldn't be a truth that needs exposing.
have you met Americans? internalizing uncomfortable truths is not the strong point among most
This is the most measured take on Chomsky I've come across since the Epstein revelations. I think you're right that his split personality regarding DoD money is a lot more important to viewing what he did (as well as what he didn't do!) than the mentions in the Epstein files, which seem like small beer. He (or his wife!!!) should have known better, and describing the MeToo movement as hysteria was dreadful, saying ". . . [it] has reached the point that even questioning a charge is a crime worse than murder"—What?!?! Horrible, but the reach of his expression of that sentiment is (thankfully!) limited. But by word and deed at the start of his career, he definitely normalized living off DoD money, and that's a problem we're still living with today.
Yep, best to never have heroes.
Norman declined any introductions.