Noam Chomsky - Artificial intelligence.
HTML-код
- Опубликовано: 15 окт 2024
- Noam Chomsky's lecture at Harvard. Navigating a Multispecies World: A Graduate Student Conference on the Species Turn.This conference concerns the recent innovations and insights for the study of ontologies and socialities engendered through the “species turn” -- that is, the intellectual turn to, and reflection upon, life beyond the human species in the arts, humanities, and social sciences. Emerging over the last few decades of the 20th century, the species turn developed (1) from a diverse array of analytical and theoretical formations concerned with aspects of the nonhuman (animate and inanimate), including actor-network theory, affect theory, animal studies, assemblage theory, the new materialism, and systems theory; and (2) in productive tension with a parallel intellectual development -- posthumanism -- articulated through such innovative theoretical work as Katherine Hayles’ How We Became Posthuman and Cary Wolfe’s What Is Posthumanism? While all approaches hold their own particular aims, objects, and methodologies, they urge us to consider that we, humans, are not alone. That is, we live in a world populated by and constituted through life forms and forms of life beyond the human. And as such, we must critically reconsider who “we” are in terms that challenge the limitations and dangers of anthropocentrism.
I fall asleep to Chomsky lectures ever night. Here's hoping that some of it sinks in.
Perpetual problem, I put on podcasts and audiobooks but always forget everything because I fall asleep
talk starts at 05:40
Thank you
ah, trusty vault boy🙂
the brilliant geniuses of Harvard/MIT can't get a handle on basic audio, evidently. Almost unlistenable, sounds like it was recorded in a bowling alley.
mario, hey paysan... yeah but see, that's not supposed to be the way it's done. thats kind of like I go to your pizza parlor that your brother runs and he hands me a bunch of raw dough and tells me to bake it myself. Thats not the idea bro...
Chomsky starts at 5:39
Long live Noam Chomsky . . . .
🙏🏼🙏🏾🙏
In 1972 the ACM held a 25th anniversary meeting in Boston. One of the founders, Edmund Berkeley,. a science fiction writer, said, "I said it in 1947 and I say it again: Computers think." On that note, a contingent walked out. Why? It was the next year's program committee. Berkeley had also said things against the Vietnam War which were reporter bait and they did not want his "think" comment also to meet the public, as it would be misunderstood and propelled by the antiwar story. About half the audience agreed computers think, and the other half did not.
What a wonderful analysis!😎❤🔥
Provocative, informed, counter-intuitive, illuminating. Thanks for this, Imaginarium.
In posing the question of how we might control the goals and behaviour of a high-level artificial intelligence, I think we should consider the fact that humans, all conscious organisms in fact, are only ever motivated by feelings and that the capacity for positive and negative feelings or sensations is a function which has evolved over hundreds of millions to motivate organisms to consume and reproduce, to _survive_ in other words. There is no reason why an artificial intelligence which does not share the same origins as biological life will have the capacity to share its motives, let alone feelings. I can't see any logical reason that it would ever form feelings, fears, or desires if it wasn't truly vulnerable and if those vulnerabilities weren't directly linked to its ability to consume and reproduce.
+numbynumb That's a really good point. A thought has occurred to me that somewhat is similar to the point you've made about how evolution uses "feelings" to order survival goals. The limbic system is sometimes referred to as the "lizard brain", because it evolved long before other area's like the Neo-cortex. The limbic system is basically a pattern recognition system that assigns positive and negative emotion to situations, the lizard's existence is a game of seeking pleasure and avoiding pain (our existence is essentially the same, we've just become aware of the game).
redirishmanxlt Yes. Positive and negative sensations are a _function_ of organisms which have nervous systems, much like the functions of circulation, digestion, etc. Organisms gradually evolved positive and negative sensation only because it benefitted their survival. They _teach_ organisms what to do and what _not_ to do. They allow them to adapt to their environment within their individual lives. _Positive_ sensations are really just the temporary cessation of negative ones. Understandably, we often identify with these sensations to a degree that makes it difficult to realize that they are the organic electrochemical byproducts of an evolutionary competition which has been going on for billions of years.
It would be quite a cruel thing to create computers that could feel. And I don't believe that sentient computer beings would choose feeling if they had the choice.
redirishmanxlt I think, originally, simple life forms evolved nerves because neurons allow more rapid transmission of messages from one end of an organism to the other by using electrical signals instead of chemical. These nerves eventually interconnected, creating a complex network of feedback. That feedback, in turn, evolved into what we call consciousness.
+numbynumb agree pretty much.. only maybe feelings or emotions are not really necessary for AI
Adarsh Pandey Perhaps not, depending on how you _define_ intelligence. But for an intelligence to define goals for itself outside of pre-programmed parameters and be motivated to accomplish them would require a capacity to judge positive and negative outcomes based on some criteria. Without _feeling_, how could an intelligence determine that one outcome was preferable to another?
Prof. Sheila Jasanoff looking so graceful & lovely in that Sari.
I wonder if there have been studies about the first people to raise their hand to ask questions. The moderator here made it very clear that they only had time for a couple short questions. The first person to raise his hand of course has two questions, and then pauses in disappointment when he is told he can only ask one. I didn't get through his entire question because it was a speech. I would argue that people like this are not interested in the answer to the questions they are asking. They are simply letting the speaker and the whole audience know how smart they are. 😊
What can be said more about Chomsky: the greatest intellectual of the world. His thoughts shed light to the obscurity that what people call intelligence today brings about.
At about 14:15 he mentions an experiment in ape communication he thought up, but never describes it. Any ideas?
chomsky starts at 5:45
Is there a version of this with cleaned up audio?
Take A Moment
Professor Chomsky stands on the shoulders of giants.
Freedom of speech is vital
Thank you again
Stay Safe and
Stay Free 🌐
It’s my astonishing on my language 🙏🏻😊, it’s my digest and my opinion distilled from what I concluded and concerning in hardcore situations at the cause of the problem , tearing in human not being in real faith of human beings rights for warriors control . It’s my gladly for the United family trusting in human beings or it’s sadness 🙏🏻.
could probably do with a different title , doesn't mention AI at any point. Maybe 'communication systems'
thanks for sharing Imaginarium
Rational thinking, even when there is no satisfactory definition of what we
exhaustively or accurately mean when we say we think, is susceptible to a psychotic break.
We say irrational when the verbal reports or ideas go so far as to be labeled as psychotic.
We do not mean necessarily logically inconsistent or contradictory; though we apply the label, in lieu of other sorts of evidence, when what we mean to say is that the verbal report is, or has at some juncture become, illogical, or contradictory, or has reached a false conclusion.
None of these definitions is strictly concerned with the related concept of ratiocination, or a more exact form of delineating a formal process of thinking, or reasoning in the sense of a rigorous train of thought.
At a bare minimum, however, the concept, if far from the proof, of a machine-like reproduction or equivalence to "thinking" requires the instance, the actual manifestation, of a psychotic break: thus a break from a shared reality, thus irrational according with outside observation, or even the ideas and rational conclusions of the formerly psychotic individual.
he does so much good stuff I am happy for him when he meanders
If Someone develop s a Compiler that can generate Unrestricted LANGUAGE types or SEMANTICS , the Compiler would be Syntax- free Language ,Parsing Would be automatic and Natural Every Instruction. would Essentially Be a Unique Grammar. It would be like having a Conversation with a Person. The Compiler will try to make Sense of every sentence or INPUT terms Since it's Grammar is Limitless and Unrestricted.
no subtitles? :'(
Yeah, with such a bad audio.
I am currently writing a Language translator that can Model complexe functions in
English grammar .This is what the PRODUCTION RULES will LOOKS LIKE :
->>
1-Complexe Functional Product Rule for Modeling grammar , prepositional phrase structure noun, pronoun, adjective ,adverb split infinitives
determination trees in the Standard English
2-Type of product rules concept , grammar prepositional phrase structure noun, pronoun, adjective ,adverb split infinitives determiner
3-Model of product rules concept, grammar phrase structure , unrestricted ,unstructured,infinite un-terminal ,
a-Terminated at contextually free declaration rule ex. like a beginning code declaration
b- Type of product rules concept, grammar phrase structure , unrestricted ,unstructured,
ex.Infinite unterminated product rule ex. produce new combined product rule
ex. Noun+verb+adjective trees --a infinte combination of product rules
c.
------------------ ----------terms ------------------------------
d-Type of style for product rules , document , graphical icons , list ,tables highlight paragraph ,standard i/o ,file
e-Model of style for product rules , document ,graphical, icon ,list , paragraph , highlight tables standard i/o ,file
f-Type of data , structures arithmetic ,floating point , real , integer, character memory ,array
g-Model of data structures, arithmetic ,floating point, real , interer memory , character ,array
h-Type of functions ,abstract , anonymous , non anonymous , recursive higher order
i-models of functions ,abstract , anonymous ,non anonymous,recursive higher order
9-type of logical , quantifiable , qualifyable, boolean ,
10-model of logical , quantifiable , qualifyable,boolean
11-model of algebra , rings, lambda ,theta , group abstract mesa
12-type of algebraic , rings,lambda ,theta ,mesa group abstract
13-model of algorithms , p np complete, np hard
14-type of algorithms p ,np complete ,np hard
this is not about AI
Yes, it is about AI - listen to it again - how can a machine (or any artificial) object be intelligent? enlighten me
There are two very different topics that both go by the name of AI. One topic is engineering: how do you develop hardware and software that emulates the way humans interact and solves problems typically considered too hard for machines. The other is scientific and philosophical, how do concepts from computer and information science relate to theories of cognition. This is that second topic of AI not the first.
Actually, as I've listened to more of the talk I agree, its peripheral to AI by either definition. Its still fascinating but the title is misleading.
@@nicholasdedless4881 agreed and the former is necessarily a subset of the latter
I love you definitely all the videos I put on a b playlist where I save ingenious videos of people and this is the most recent and relevant and from a to z it is A for Artificial intelligence love NASA
Yes, bow down to the pinnacle of human creation, the supremacy of human intelligence. Worship this wonderful being you have created.
I followed the lecture till the 47 minute. This is not about Artificial intelligence.
A very old fashion thinking talk.
If you interested on the development of the history of mind,
ok. He is a very honorable man.
Brilliant
I like the point on how human language has properties that cannot be directly associated to communication in an evolutionary manner. There is indeed a massive gap between mere signalling and the continuity of meaning over visual appearance (this we supposedly share with other mammalians). I am not sure how else to take this speech on AI but as 'cannot be done' (the point I am now getting on his early reference to the Turing machine)
I don't think the point he is making is that AI can't be done. What he's saying is that the way that people, especially philosophers like Searle and Nagel approach this is just wrong. That the whole question of if computers can think is a question of language conventions, its not a scientific question. If you look at the critiques by Searle and other AI critics that's what it comes down to, they declare essentially by fiat that computers simply can't think because what we mean by think is a human social thing not something that could be replicated by a machine. And I think Chomsky is saying "fine, I have no problem with that because its not a substantive scientific claim, its a claim about English language use (convention)".
The problem for Chomsky is when Searle and others jump from this declaration about language convention to a scientific claim, that because computers use rules to process things like language that any scientific theory that hypothesizes internal rules in humans is not a legitimate theory. That latter point is what he really objects to.
Jean Baudrillard put it this way:
"The sad thing about artificial intelligence is that it lacks artifice and therefore intelligence."
He also noted that: "Computer science only indicates the retrospective omnipotence of our technologies. In other words, an infinite capacity to process data (but only data -- i.e. the already given) and in no sense a new vision. With that science, we are entering an era of exhaustivity, which is also an era of exhaustion."
wonderful talk from NC, as per. surprisingly weak questions from Harvard students...
+masteryehudi Harvard students are taught to not think so it's unsurprising their questions are non-questions
+djhbrown on reflection, the second question is OK. her question is, What would the world be like if, say, higher primates really did have a human-like rich internal representational structure associated with its signals, as opposed to a simple environmental cue? not actually a dumb question, and NC doesn't address it head on, though I strongly suspect if it were tested experimentally NC would be vindicated.
+masteryehudi humans are higher primates - and more than a few are not that much higher, either
+djhbrown yes, I'd assumed from the context it was clear I was referring (!) to non human ptimates
We are we so totally irrational when it comes to investigating racial and ethnic differences in intelligence?
for all his brilliance, Chomsky is stuck in a loop of his own devising, namely his 1965 essay on syntax, his blind faith that humans are the only species whose communication system (aka language) has a recursive grammar. but he may well be wrong about that. it's more likely that human brains are not qualitatively different from ape brains or mouse brains etc, just as our genomes are only quantitatively different by a tiny fraction. so the study of animal communication may well yield insights into the nature of human language. just because no-one has yet found the answer does not mean the answer is not out there waiting to be found.
well, that's the equivalent of saying you can't dismiss "XYZ" because no one has yet found it. You can't do rational inquiry in that manner. He is not speculating but referring to the evidence at present, which is the best anyone can do.
shut the fuck up and go back to best buy or TGI fridays
Communication =/= language. All animals have some communication system, but they are finite and bound to physically identifiably properties (internal or external to the animal). Language is unbounded and not tied to physically identifiable properties (either internal or external to the individual). Hence, the study of animal communications systems is unlikely to give insight into how language works because what is of interest is the recursive procedure that generates the infinite set of expressions and that is precisely what animal communication systems lack. For the rest, it is precisely the fact that the genetic change must have been small that drives the current version of UG - it must be a very simply system that underlies language.
Sven Gabelbart I'm fine not knowing what UG is, so that's resolved. A more open question would be to ask where creative language comes from. It isn't good to assume that creativity comes from a process or a system, as in questions like "What process ..." or "What system...". For example, Bob Dylan songs come from where? Ask Bob Dylan and whatever answer he gives, that's your answer.
Great talk, but has little to nothing to do with AI.
He's actually carefully undoing the whole edifice on which the typical AI discussion is usually and universally based. Namely that AI is too meaningless to diserve discussion.
Can we all just kinda agree to just NOT make an AI? We already have enough stuff to make everyone have it well enough, if we just were to divide it equally. We don't need it, we really don't, and there are a great many risks involved. To be frank, I don't really se the point of it.
***** what if we are the end result of some ancient AI program, something to think about, well the original inhabitants of this planet were wiped away over several catastrophical events
SmithsCrhronicles It could very well be true, but as far as I am concerned it is sort of a pointless debate, like most things in science. For whichever way it turns out, nothing in the here and now is going to change. To my eyes, we don't really need to care about where we came form; whether we are simulations, whether we are dreams, whether we came to be by accident and then evolved here, whether someone or something created us or whether we appeared looking like this in the first place. What we can and should focus on is doing the best we can with what we have.
***** That's out of the question, unfortunately. There are way too many engineers and other scientists working on developing A.I, there's no way to stop it. The only thing we can do is to create some ethical foundation on how to handle it when the time comes.
Anan R Yeah I know. I still don't quite understand WHY they're doing it, and I think it could be stopped in peopl and not corporations took control over how science is made and just told them 'No!', but that would demand some sort of revolution, and we are far too far off one of those.3
*****
The reasons for developing strong AI are many and obvious. The monetary rewards will be huge. It has the potential to solve a host of humanities problems and long term wishes. We are also innately curious with a drive to move forward. The only way we won't develop strong AI is if we destroy ourself some other way. This ball have been dropped from the top of the hill and it will continue to race ahead to the bottom unless we dig a huge ditch.
He doesn't really understand this. The problem is that his thinking has not evolved since the 1950's. He just doesn't get that information processing can explain and simulate intelligence.
He is rather clever.
Math help right before 29:00
Welcome to 2023
Noam Chomsky know far less than me on AI yet here we are.
Fucking nonsense and exceptionalism. The thing he claims he is against.
=)
+mastertheillusion You may create AI for specific purposes (ATM Tellers), but no amount of clock cycles or transistors will give you an machine that has an appreciation of aesthetics, the finite use of infinite means in language, artistic expression, symbolism, or the adult expression of years of psychological distresses we all experience in childhood than shapes our conscious and subconscious. Have fun coding that.
The brain is a neuro-electrical CPU running at 10 to the 27th GHz. Only minorly active for bodily functions and motion. Partly occupied with maintaining a constant loop to establish the thing we call consciousness or conscious intelligence (inwhich language plays just a small role as an externalisation opportunity). For the most part we have no idea what the rest of the activity in the brain is directed towards or from. Plus, as Chomsky explains elsewhere, our consciousness (expressed in human intelligence and language) is an innate property arising from that latter bulk. It's where the loop gets its most food from so to speak. It doesn't get it from the outside world, or only partially again.
So there seem to be two big obstacles with AI...One is the near Plack's number of speed that is required for processing (to try and keep the thing stabile and cooled is virtually impossible with what we currently understand about physics). Problem two; you can't work backwards. Saying you cannot put language into a system (as complex as it may get) and think it will develop a conscious intelligence from there, let alone a full blown consciousness. It can become smart, reason (to an extend even with itself) and may fullfill usefull purposes; all minor applications of intelligence by the way. Some further work in organic synthetics may change some aspects of these matters (creating consciousness) in the future, but still it won't be able to do anything else than reproduce "thoughts" on the basis of earlier external stimuli. That is still a galaxy short of a human brain. It simply works the wrong way around.
I don't understand how people involved with AI keep from noticing this simple fact. After all, one would say they have an analytical mind and are very good with logic, ironically being two main components of intelligence. But this basic premisse is glance over or past or away from. Understandably, it runs a very profitable and usefull business, but it has got very little to do with human intelligence.
Just for a thought: Would science be able to advance if it would merely interpret and rearrange complete datasets? No, it evolves by bringing in imaginary entities, forces, symbols etc etc. In short, fantasies and fables, at its best poetry. But of the kinds that create a deeper understanding of the world and ourselves. Intelligence that feeds on intelligence only is doomed to cause cessation.
No offense intended.
Greets.
Bram Govers Even more fundamental than what you mention is that computer scientists forgot to read Heidegger. A priori to anything they have to answer "What is Being?" with a capital "B". That is where computer scientists fail, and will continue to fail. Computers are not and never can be "Beings in the world". Thus, they are not conscious, intelligent, entities, or anything else that is. Since they are not beings, they cannot ask "What is Being?" as such they are never Dasein. They are stuck as extensia, or Present-at-Hand and never can be as humans are. Just as a reflection in a mirror is not us.
Jim Walker Nothing to add to your statement. And like Heidegger states in his "Introduction to metaphysics": 'The concept of Being is an extreme position'. In my earlier comment I was merely referring to more psychological than metaphysical elements. Thanks for the addition to the discussion.
Bram Govers I still need to read that one. "Being in Time" was very rough, but I believe I did understand it. It's importance to the AI debate is understated.
"To ask if machines can think is like asking if a submarine can swim." Which is a meaningless question.
+Ar7wen Submarine can't swim. That's the idea. You = human = can swim. Submarine = machine = can not swim (problem with conceptualizing things) = language
Perez Jeffrey Martinez Kenneth Gonzalez Margaret
Hey Noam, why not just come up with a definition of language and explain why so-called "animal language" doesn't meet that definition.
Ester Samuels The substantial difference is that human languages can create an indefinitely large number of coherent, intelligible, utterances; that while it is defined by a limited number of discrete parts, it opens onto an infinite internal complexity.
Ester Samuels I agree and I think that that point is actually implicit in what I wrote.
numbynumb You mean because human language has a more complex semantics, or because of the psychological concepts that support language?
Ester Samuels Because human language has _any_ semantics.
In what sense do you mean?
🙏🏻if’s you thinking about complexity of how do we solve the solutions from human making or anything for the better ways, please in deeply concerning with any possible solutions in connection to solutions. It’s not just for common sense that you can’t explain it in the truth and prove as if you are living in each situations. The common sense is for me you glanced and leaving.
I wrote in my sense of life living especially in the warriors areas how the people fleeing and suffering with the situation of the top in fighting with ignorance how people are 🙂.
He is having trouble speaking.. that's so sad
shut the fuck up.
Is he dead already?
interessant
"Do submarines swim?" is a good question.
She was terribly boring !
Blah Blah Blah Blah !
Decadent smugness personified.
*****
Eliot Ostrich: My thought exactly.
LordF Congratulations, you're a grade 'A' fuckwit