Noam Chomsky on Artificial Intelligence, Language and Cognition

so um welcome to the east of philosophy and Technology it is indeed a great honor for our Institute and dar panel of researchers fellows and Scholars to have you in this philosophy discussion about artificial intelligence language and cognition and I would like to thank also our audience for their patience and enthusiastically joined our event today there are too many people here is an eminent philosopher linguist and political activist MIT professor emeritus and Laureate professor of the University of Arizona author of more than 100 books and he's one of the most important intellectuals and thinkers of our time due to remarkable contribution in modern Linguistics philosophy and his active role at the social critic Professor Chomsky I must admit that since my work is on platinous philosophy I am tempted to ask you about some neoplatonic elements of your philosophy and they came with neocolatonist but I will remain focused on our topic today and I know that my question sounds generic but I would like to ask you if you see any implications of the rapid developments of artificial intelligence on the growth of human language faculty you have expressed your skepticism about new trends of AI mainly on Advanced statistics and Big Data and my question is to what extent the growth of the human language capacity particularly the state of the language of of the knowledge of language might be affected by the rapid development of AI well the quick answer is not an old not at all there's no connection and there's a very simple reason for it the approaches that are being used the large language models being based on deep learning approaches are designed in such a way that in principle they can tell us nothing about language or cognition in fact they are designed in such a way that the better they work the deeper are their inadequacies and there's a very simple reason for that uh can explain with an analogy uh we're all familiar with the periodic table of elements suppose somebody comes along with the better periodic table it includes all the elements that exist or the ones that can possibly exist and all the ones that cannot possibly exist and it makes no distinction among them so it doesn't distinguish between what can be an element and what can't be an element of prevention this system is telling us nothing about elements now let's turn to large language models say gpt3 or anyone else you like they make they what they do is scan a astronomical amount of data find some regularities and string together things that conform to those regularities the work exactly as well for impossible languages as for possible ones put together any text like of astronomical size with completely random Properties or properties that are designed to violate the rules of language the gpt3 will do fine doesn't make any distinction so by definition it's telling us cannot tell us it's not that it doesn't but cannot tell us anything about language or cognition it's a whatever its uses you can ask but it just simply has no relation to any question having to do with uh science or philosophy it simply basically it's High type plagiarism and it'll do the same plagiarism with impossible languages and again the better it gets the worse it is because the better it gets at another trillion parameters and you know double the number of terabytes you're using it'll do better on Impossible languages so it's worse and it's an impermeable problem there's a lot of critique of these systems huge literature in fact showing things they get wrong about language well that's interesting but that can be overcome by expanding the system but you can't do anything about the fact that in principle [Music] so what you can ask about these systems is do they have any utility in engineering advantages there are AI systems that do that in fact I'm using one right now since I don't hear very well I'm reading transcriptions well the transcriptions are produced by something like large language models so that's a useful application but it's of absolutely no scientific interest doesn't tell me anything about how humans hear or perceive or anything else I thank you very much and and you mentioned uh actually that GPT in this new open AI systems and there is a question about that from Professor Dukakis at the onion University and he would like to ask you about this issue because of his of his research interest uh welcome Spiros to thank a professor [Music] let me take you through the field of Education from private school to his best in order to discuss the role of artificial intelligence and other Technologies in enhancing and achieving learning recently we saw an article published by an academic in the chartered GPT at the same time New York City's Department of Education announced the ban before chapter from schools devices and Network what do you think the future of Education will be and how will be affected by artificial intelligence machine learning and models like such gbt and how can educational policy medication teachers act on this issues to strengthen the new generation and moreover do you think that portability devices and other digital biomarket measurements technologies have a role in education and can they contribute to improving Clinic with the use of AI thank you well I think we can remember that these AI systems are basically plagiarism high-tech plagiarism so the question we can ask is how does plagiarism affect educational policies well it does for years teachers and professors in social sciences Humanities have had the problem of providing students with a ask them to write an essay on something and they'll hand in an essay which was plagiarized from something and there are plagiarism detectors that are available to faculty to ask to help them check whether the essay was real or plagiarized well now it's just more advanced plagiarism so it's a bigger problem a student can uh is supposed to write an essay on some topic they can play around with the chat GPT and it'll produce something that looks like an essay on that topic the way it's doing it is by plagiarizing materials that have been written on that topic just doing it in a more fancy way than picking it out so that's a problem and I think that's why there are no efforts you saw the New York Times article to ban uh the devices for students or to find other maybe require that students hand in essays by handwriting or redevice the notion of essays so that you can tell if there's an individual element and it takes a little work but can be done what he you know and uh basically these are just impediments to education they have no educational function their impediments the proper education uh students asked to write an essay on some topic instead of thinking about it and getting some ideas and working them out the student can plagiarize okay that's basically what the problem is we might ask whether these systems like chat GPT have any function whatsoever in a useful function I mean are they like life transcription or Google Translate which is useful you know you want a rough if I want a rough idea of what some article in Greek is about I can use Google Translate it'll give me some rough idea of what the article is of course I won't trust the translation if I want to really understand it but you get some kind of hint is it worth going into further or something like that well that can be useful but is chat GPT useful for anything I can't think of anything it seems to me uh just a toy that's used to mislead people [Music] I mean you can understand what young people like it you have this huge computational capacity at your your fingertips it's like having a wonderful game you can play with but uh it can't contribute to any question of science or understanding but in principle and it doesn't seem to have any utility except the negative utility of making it harder for educational procedures to work properly so I I think it's uh as far as I can see it's a net loss thank you very much foreign would you like to try turning on your camera perhaps now it's YouTube works I don't know if you if you wish I don't see anything and camera on [Music] okay well that's not doing anything thank you thank you very much now we can see you clearly uh we have a yes and all these things are extremely interesting and fascinating for us um there is a question from our partner in at London um Petros La fazanidis who would like to ask you about cognition and the role of cognition hello Pedro from Athens hello Professor an honor to talk to you um so I I had a few questions and I think they kind of been answered so I was wondering whether I could change it a little bit and um is it about cognition so um what is your current view of cognitive Sciences as a field and in the past interview you have mentioned that it is almost a before Galileo stage I'm wondering whether this view has changed and whether you have you finding something that is some ideas or some people that you find interesting and promising in contemporary cognitive science that's right yes there's a lot of quite interesting work so for example in one of the recent issues of the journal cognitive psychology there's a joint paper by Stan dehane of the kalistophons and a number of others in which they investigate pre a proto-human uh symbolism the use of Jim from artifact archaeological artifacts they study the geometrical objects that are used that were used in a couple hundred thousand years ago and it turns out they're interesting regularities they have a nice theory that explains them and suggests and have some with contemporary studies of humans looking at these things what parts of the brain are involved and it I think the kind of overstate the evidence but the results are I think to show that proto-humans early before modern humans had some form of decoration decoration isn't art there's a distinguish between them so if you look at a yeah painting in the museum [Music] painting of Madonna by Raphael there's a lot of decoration around the painting but the decoration is very different from the painting the painting evokes thought understanding uh you know exploration and so on the decoration around the painting just makes it it's interesting that it's nice that it's there it's better than a blank space but it doesn't have the effect of the painting and the art historians have made this distinction and discussed it in fact gombridge and others and I think what the hand and his group are finding is that prior to the appearance of modern humans there was a sense of decoration ways of organizing geometric geometrical shapes I can't draw them here but you know like an angle or something or a lion a chip on a bone and doing them in ways which amount to decoration chimpanzees they sure can't do this just kind of interesting uh so there was something in early modern human before modern humans that had already developed some conception of uh decoration that is not that other primates don't have and that maybe some kind of precursor to the cognitive capacities that humans later developed although they also showed that different brain areas are involved well I just take that as one example because I have to read it a couple days ago but there's a if you look through the cognitive science literature there you find things of interest some of them are it's very hard to as the main topics that are studied in cognitive science seriously or vision and language in fact Jerry fodder once said as a sort of a joke that cognitive science equals vision and language well those are the two topics you can study in some depth and there is interesting work but uh and there's a very striking difference between the way they're studied in the case of vision human Vision we know quite a lot about it the neurophysiology of it and so on a great deal is understood about language you know I understand almost nothing about the neurophysiology some things vaguely about localization not much more and there's a reason for that human language is isolated there's no analogs in other species so you can't do comparative work and you can't do experiment at ethical requirements bore invasive experiments with humans we know a lot about human Vision because we have about the same visual system as cats and monkeys and rightly or wrongly we do do invasive experiments with cats and monkeys like raised kittens in without under various conditions of visual deprivation that teaches you a lot about the nature of vision facting classic work by David hubel and torben Visa Nobel Prize laureates it was based a lot of it was based on experiments with kittens raised under different conditions of deprivation visual deprivation along with the invasive experimentation like putting on electrode into the visual parts of the visual cortex and so on we do all those things with animals you can argue about whether it's right or wrong but it's done and given that they have the same visual systems essentially we learn about human vision but you can't do that with language there are no other organisms it's totally isolated there's no analogous systems anywhere and we don't permit the obvious experimentation with infants like raise infants under different conditions we're not going to do that so therefore to try to discover the neurophysiology of language is a very difficult task there are some ways to do it with non-invasive techniques uh a recording of the scalp [Music] FM or fmri you know things which give you an indication of something that's going on in the brain but it's you can think of all sorts of experiments that if you were say mangala you could carry out and learn about human language but we fortunately we're not all manglers so it's a very hard topic thank you very much thank you professional tomski are really exciting all these things and one of our fellows Christus Jonas who is now in Sweden I would like to ask a question now so uh Crystal would you like to ask professor samski yes so excited very excited to be here and to meet you in person um a question I would like to ask uh I have I wrote a lot of questions but uh I'll I'll start with I have two questions so I'll start with one um one is more about conditioning the other one is a bit more questions so the first one about position is embroidery for the mission has been seen its tremendous interest in recent decades such leaders have in many respects diverse from single purely computational understanding of coordination in the mind do you think theories you can go into the mission will come to inform current practices in artificial intelligence or in general in our fears about the Mind who called the group science and then make a contribution to understanding uh from God Yeah in our understanding of language understanding well there's certainly there's not I mean it could in principle undoubtedly error use of language involves more than just what's above the neck you know other things are going and when we're communicating we're doing it involve all sorts of things are involved there are emotional state the way we present ourselves with gestures a lot of things are happening in our body that are not just brain activities that are involved in the use of and they use certainly in the use of language and acquisition of language and so on so in principle yes it's possible then we have to look and see uh well when the work is done what do we learn uh it's not easy to find much in my opinion but it's certainly a field that it's sensible to investigate and do you think would you say that would it be considerable that the important cognition could also inform our current artificial intelligence model in the future or our ability to understand the intelligence and model intelligence yeah thank you well oh the goals of artificial intelligence or to try to find out how a human being or some other form of intelligence can function in the complex world so if you read the so-called theorists of modern artificial intelligence the you've upon others they say well we're not really interested in these narrow questions like how does language work or how does vision work but in the broader question of how does a human being function in the complex world well that's a typical too much too many variables question I mean there's a reason why scientists don't if they want to study the laws of motion they don't take uh astronomical numbers of videotapes of leaves blowing in the wind yes that's interesting question how do leaves blow in the wind but you're not going to find out the laws of motion that way I mean you may get some statistical approximation to something that's going on they find regularities but it's not the way scientists studied the laws of motion so you mentioned Galileo I had a Galileo study the laws of motion with thought experiments into the experiments most of the experiments he describes couldn't have possibly done think it through uh sometimes you can do an actual experiment based on what you've thought through but in any experiment is a radical idealization any just any experiment you do in a chemistry lab or anywhere else is radically abstracting from the complexity of life and experience and there's a good reason for that if you want to find out the basic principles by which the world works there's no point at looking at things that are happening in the world just too many variables involved so what you do in experimentation thought experiments is tried to select the things that you think are important and throw at everything else that's radical abstraction and thought experiments go on right until today like take the major achievements of 20th century science Einstein's relativity Theory quantum theory a lot of it is just based on thought experiments like uh relativity theory was based on Einstein's uh asking uh what would the world look like if you're traveling at the speed speed of light well he wasn't traveling at the speed of light but you could think through what would happen what time would be like and so on if you were or take the famous experiment with the is the cat alive or death dead depending on how you're looking at a particle moving through a screen I don't know if any was actually bothered with the experiment but the the whole discussion debate and development of it was based on the thought experiment so we carry out thought experiments today too sometimes you can carry out real experiments but they're all radical abstractions well the kind of approaching artificial intelligence which says we're not interested in these special special systems that have their own rules we want to deal with the whole matter of how humans navigate in the complex world how are you and I doing what we're doing now I don't think anything is going to be discovered that way just as nothing would be discovered about the physical laws if we just took the world as it is and didn't abstract radically from it so my guess is that the progress of artificial intelligence will be to study specific systems that we can isolate abstract uh when you're studying say Vision again go back to the ability when they were carrying out their experiments they didn't study how a cat develops how a kitten develops the the imposed sharp restrictions like a kitten which is exposed only to straight lines the horizontal lines let's say what happens to the visual system then they're very radical abstraction from experience and then you can learn something if you just said let's take a kitten and take all the environment around it and let's see what happens in the brain you'd find nothing there's just too many things going on and I think that ideal for artificial intelligence that well in principle you know you might must something might happen it doesn't look like a very sensible way to discover the nature of what's going on thank you very much so uh I would like to just ask one more question which is uh uh what would you advise a young student looking for purpose and aspiring to grow their understanding and intellect to its fullest potential well I think it's uh we have from infancy a kind of a natural inclination to inquire to understand to find explanations anyone who deals with young children say a parent knows that the child is driving you crazy by asking questions why does this happen this way why does it happen that way you try to shut them up after a while because it's too annoying children just want to make sense of the world want to understand how things are working every human society that we know anything about has asked similar questions what's going on with these objects up in the sky how did how was human life created and it's done with myths with religion and more sophisticated ways it becomes science so I think for a student a young person follow your natural instinct of trying to understand a lot of Education those that instinct a lot of education is designed so as to prevent people from doing it like the dominant educational practice in the United States I don't know about Greece but in the last 20 or 30 years has been the principle of teaching to test you have to rigorously teach students schools skills material which they learn and then regurgitate in tests absolutely the worst form of possible education it was ridiculed during the Enlightenment centuries ago it was described as the model they used is pouring water into a vessel and then letting some of it come out the worst possible form of Education that we've all had experiences with that I'm sure I'm sure you have you have to take a certain course you don't you're not interested in it but you have to get a good Mark so a week before the exam you study hard and learn a lot of stuff and answer the questions in the exam you get a good Mark two weeks later you forgot what the course was about uh we've all had that experience in fact a lot of education is like that the worst possible way the white way is for education to stimulate and encourage the natural inclinations to understand inquire create and so on this is just instinctive to us so encouraged and stimulated it's a you don't get good metrics that way you don't get numbers that tell you how well the student is doing if you want to automate the educational process so that you can tell whether this school is meeting some kind of Standards you know then do the worst kind of education do teaching to test and then you get the numbers which are mostly meaningless but the stimulation and that's real education I remember for example myself one of the best professors I had in graduate school mathematics professor oh he'd come in we he would come into class there was something we were supposed to read and study he'd come into class and write something on the Blackboard those days they had blackboards and chalk I write something on the Blackboard and say is this a theorem and then the rest of the class would would be devoted to trying to figure out if that's a theorem or not it wasn't I mean he would intervene you know there was some direction and shaping but basically it was students seeking ways to determine whether this kind of this statement is or isn't a theorem how would you go about it that's education that happened to be graduate school but you can do it in kindergarten they're a very good science courses programs which try to to stimulate such understanding so here's one that's used known bitter schools they take kindergarten children and have it and give each child a shell and on the show put some objects like a button a stone a seed a bunch of other things and then you have a question which of these things is going to grow that's the question then the children have what they call a scientific conference in which they try to figure out things they could do to determine which one could grow in the various ideas are proposed and then they try them and teacher kind of shapes it a little bit but mainly they're stunning finally they figure out with experiments that it's a seed that grows when they have gotten to that point uh each child is given a microscope teacher splits the seat open they look inside and see what is causing the seed to grow they're excited interested and so on they learn something well that's education take another example from I think this is around Middle School in the town where I live there's a section of a very poor section mostly immigrants Mexican immigrants very low educational level parents barely speak English often kids go to school schools are underfunded so one teacher may have 50 kids in a class and school is mainly a problem of keeping the kids from hitting each other you know nobody ever went to college even high school well a couple of teachers and some University faculty decided to run a experimental program what they did was put in a small garden in the school and they got the kids too water the garden and get things to grow and sort of fun that then they finally they brought in a fish tank and the fish tank and they began to study the excrement from the fish tank can be used to fertilize the garden well a couple of years later I went to visit the place we were brought a couple of young kids 10 11 year old kids took it they had this very elaborate Garden fish and all sorts of things they took us through it they described to us the chemistry and the physiology of how things grew and how well the nutrients work and what happens to the soil and so on they had learned science no discipline problems anymore as they go to high school and college they've integrated the neighborhood into it a lot of the neighbors the people in the community are a generation away from being composinos in Mexico they know about agriculture so they are interested they started selling produce from the from their own garden in the neighborhood the community a community developed a participation well that's education you can do it in kindergarten you can do it in Middle Schools you can do it in graduate school different ways but the worst kind of education is study pass the test forget about it and there's much too much of that and I think it has to do with basic features of human nature that we see in from infancy with just the desire to understand what this complex world is about well when you organize and direct that instinct properly you get good education thank you very much we have another student partner canelopoulos who would like to ask you a question uh Pano good evening professor uh uh it's just an honor to meet you what do you think about the theory of some people that creativity is a privilege of human intelligence only and not the artificial one in the distant future would you see a Nobel Prize in literature in the hands of a robot thank you very much well not any first what's a robot first of all I'm in a robot is not something from Star Wars it's a computer program what we call machines or programs that's what a machine is it may be implemented by something that is built to look like a cute creature or something but it's just a program that's being implemented by some mechanical device just as in your laptop you have a calculator it's a program it's a program written in a notation in which the computer can implement it but the implementation by the computers is nothing you know it's just apart from that apart from implementing the program the computer is just a piece of metal you know that he uses a paper weight but so what we're really when we ask about robots we're asking about programs can it's going to be a program that would have creativity but what's a program programs in theory it's a strange kind of theory because a program is a theory that has to answer every question theories rarely do that like if you're even in physics you know the most advanced field they're just questions you don't know the answer to so you don't put it into the theory but if you're trying to program it you have to put it in even if you don't know the answer program doesn't work unless every question is answered so there's a special kind of unusual kind of theory which every question is answered even if you don't know the answers and it's written in a notation that can be implemented by a computer so what the rephrasing your original question in these terms we're asking can we have a theory of creativity well in principle we could but we haven't a clue how to proceed nobody knows anything about it so if somebody can figure out what creativity is and develop a theory about it then you could implemented and you could you could rewrite that theory in a notation which uh some physical object could implement but since we don't understand anything about creativity we can't say anything about it [Music] thank you very much professors thank you partner for your question and there is also a question from a final one from Noel batillo our philosopher and artist who is an ipt fellow Noah I'd like to uh ask the question you have yes hi oh I'm sorry uh president Michu Professor Chomsky um I've uh I've always wanted to meet you and hear one of your talks so this is a great opportunity uh my question is um I myself am not a computer person or a computer expert um and I don't know uh really a lot about AI um what does have a meaning for me is to know I guess in the sense that I'm not using the term artificial as in computer but wanting to know what what uh true or a real nature of intelligence is um in a human being um as opposed to you know something that's more uh artificial and fabricated so I guess my question to you is this um if you look back at your entire life and career as a linguist philosopher and political activist among so many other things um in terms of everything you've accomplished what is most meaningful you for you as a person in terms of the nature of intelligence that is not artificial but real tangible um solid meaningful purposeful um or what is the meaning for you in terms of the nature of language or of cognition well this the first paper in artificial intelligence wasn't the term wasn't used then but the paper that basically initiated the field was by the great mathematician Alan Turing who basically developed the leading figure who developed the modern theory of compute computation computability mathematician biologist his paper was called the can machines think machine and of course program kind of program think and he said this is the paper in which he developed What's called the Turing test you know which you may know about a test for machine thinking he said the question whether machines can think is too meaningless to deserve discussion and the reason is we don't understand what thinking is so he said let's see if we can find some very narrow condition which we can study and maybe that'll help us improve our understanding of machines and of software and maybe people even start using the word think to deal with this that's his so-called imitation game the Turing test where you see if you can get a program that'll fool somebody into believing that it's a person okay because it does well enough like chat GPT can it fool somebody into thinking it's a person he said okay let's study that but let's not confuse ourselves by saying we're we're studying thinking we're just trying to find out something about some aspect of what might fall under this broad concept uh touring of course didn't know it but the he was this question had been asked in the 17th century and given a sense actually the same answer Descartes famously had argued that humans have a special property race kogitones which machines don't have machines meaning old animals and most of human behavior and he had some pretty good Arguments for this one of the main ones was Yusuf language he argued that the normal use of language has a creative aspect which can't be duplicated by any any machine Any deterministic process and his associates especially Jacques Cordova asked an obvious question can we develop tests which will determine whether some other creature has this property so it takes some creature that looks like us can we determine whether it has the property risko get done it's kind of like a litmus test for acidity and the answer that Court Amwell gave was essentially tourings let's see if we can run a battery of tests ask questions see if the creature can answer them in an appropriate way if it can do it in complex situations we'll assume that it has this property of riskogadence notice that Cordon wall was asking a metaphysical question does this creature have such and such a property and remember in the 17th century that was a very serious problem because if it has this property it has a soul and it has a soul that has to be christianized then you go on from there into a pretty ugly history of colonialism but these were serious questions it's not not you know not questions of methodology and so on fundamental questions well the Turing test is like that so now we're back to you your question can we have artificial intelligence if we do it'll be a study of natural intelligence of artificial intelligence in for people like touring or the early pioneers of the field like Herbert Simon and Marvin Minsky for them it was just cognitive science can we learn something about the nature of cognition intelligence by using these methods that we have that computers and computational Theory make available and the answer is yes we can going back to your specific question in this case of language by studying computational theories of how language Works can learn quite a lot in fact I think a lot of progress has been made in these years into understanding what you and I are now doing there are a lot of questions we can't answer we can answer questions about I'm an Aristotle made a fundamental distinction between possession of knowledge and use of use of knowledge crucial possession of knowledge in the case of language is the system that's coded in your brain the sum system coded in your brain for me it's English for you it's Greek Aristotle argued that they're all the same he was probably correct and the underlying system they're probably pretty much even exactly the same for all languages he argued that they differ in the way they're externalized so you sounded out differently in Greece Greek and than you do in English but the internal system's the same it's kind of like a computer your laptop for example has a program in it and you could hook it up to one or another different kind of printer uh the lap the program doesn't care it's the same program no matter what printer you hook it up to and it may be kind of like that for us well that's the language itself then comes the and about that you can learn a lot then comes the question of use of language now we're back to Descartes how can it be that we can use language appropriately in situations construct infinitely many thoughts Express them in linguistic form then we can do it in ways which are appropriate to situations but not caused by the situations that's the basic property that led Descartes to establish the second substance race kogi duns some substance independent of mechanism which humans have that enable this to go on well about that we understand nothing at all there's nothing understood about how you and I decide to produce particular thoughts on on occasion of explicit occasions and decide how to formulate these thoughts and try to convey them to others I may go back to the 17th century these questions were regarded as some of the most important and amazing questions that arise Galileo for example was odd and Amazed by the fact that humans can somehow create infinitely many thoughts and with just a finite number of symbols can construct infinitely many thoughts and can even find ways to convey these thoughts to others so that others who have no access to our minds can understand what's going on in our minds they regarded this as one of the most amazing facts in the universe and this was not uncommon you go to the Paul well Monastery tanzanite monastery Pascal or no in other same ideas and they're oh on amazement is quite Justified and Al Galileo for example regarded the alphabet as the most amazing spectacular human invention because it could capture this astonishing capacity well come back to the couple of centuries later to where we are now we can still stare at it with oh and amazement but can't say anything about it we don't know anything about about use of language we have almost nothing to say how do you pick out a particular thought from the infinite number of thoughts that's available to you from the system that's coded in your brain zero understanding we don't even I suspect the question is probably beyond the range of human intelligence but in any event you can think you can stare at it but can't answer on the other hand if you ask what is the character of that infinite set of thoughts that you have code in your brain can say quite a lot about it uh what are they possible systems of thought non-possible ones however they constructed a lot to say about that how are they used blink wool it's a it's a more complex question of an even simpler one how do I decide whether to lift my finger can't say anything about that either that's uh you know scientists will tell you it's determined by the laws of physics but they have nothing to say about it so that's just an Article of Faith you know many months thank you Noah very much for your question uh Professor Chomsky um we are running out of time but since you mentioned Aristotle and you mentioned some key moments in the history of philosophy and science you can you tell a few words about the card worth and his influence in your theory and your philosophical background well like a lot of things it was sort of the other way around I had arrived at various ideas and started studying the early I should know that in I mean I studied philosophy at the best places Harvard University and so on but you never studied the history it just was considered irrelevant you know you don't need that old-fashioned stuff we start from now when we do everything maybe a couple of sentences from Locke or something but I got interested in the early modern philosophy because there were some interesting hints there that looked pursuing as I went on I found that a lot of what we think we're inventing today was in fact developed in quite interesting ways and throughout history back to the Greeks in fact back to Indian philosophy in fact which is an interesting tradition I don't know that much about it but in the western tradition from the Greeks up through the rediscovery of Aristotle in the 12th 13th century and Roger Bacon the near greatness who had an extremely interesting ideas about the nature of cognition and language people were very hard to read like Ralph cudworth it's almost unreadable but if you think about it there's a lot of very striking insights into their the nature of somatics and structure of language and thought anyhow I got interested in this and but it was first having sort of the basic ideas and then finding out how they were studied in quite a rich and insightful way throughout the history of thought but thank you very much uh actually there is a final question from aristea Arista would you like to to ask professor tomsky yes of course hello Professor it's such an honor to meet you and thank you for uh your answers throughout the session so well um I'm a political theorist so I'm more interested in uh social phenomena so I I'm interested in the relationship among Voice language and AI um for human beings voice is a unique characteristic that connects the Mortal body with Society through language through the the use of language from each subject and in large that's why voice is a very important concept when it comes to our understanding um of Notions of Freedom oppression democracy and so on and my question is how does the emergence of AI Technologies alter the um this relationship between voice as part of the human body and language and how would this affect social cognition and our ability to understand subjective narratives and social phenomena in general thank you that's uh extremely serious and very hard question different cultures and societies do have different ways of thinking about the world and interactions among people and so on uh it must be the case that there are fundamental uniformities there's a lot of loose talk about you read a lot of the French post structuralists Foucault and others and say well everything's just a development of culture completely makes no sense there's no way for an infant to discover the nature of the culture that they're living in unless the infant has built-in structure otherwise it's just noise you know it's just like you I happen to have a couple of dogs they live in the same culture as my children but they don't get anything you have to have something inside you fixed that's determining that this is part of the culture and this is just noise and this is something I pick up and that's something I put aside so there's got to be a common fundamental human nature that manifests itself in various societies and cultures and I think one of the really interesting things is try to find out what that fixed internal nature is that enables us even to become part of a culture which a human infant can do and it's pet dog can't do or it's pet gym Benzene it's just noise they have their own societies but not ours because they have a different nature so those are all questions that can be studied and you could learn something about them but one of the first things to do is to liberate oneself from the postmodern uh conventions about relativism they make absolutely no sense and you're not going to get anywhere by pursuing them the uh you that doesn't mean we shouldn't be interested in the differences of culture of course we should be just as a linguist is interested in the differences between Greek and English but understanding that fundamentally we're the same the language is the same they have to be and I think pretty much the same is true of cultures can artificial intelligence contribute anything to this well remember that artificial intelligence is just cognitive science using particular techniques and Technology like a theory of computation implementation of a program of theories by technology so maybe that can help but sometimes it can but you can't you can't really predict the how science and understanding will develop it's we just don't know enough to explain that so I mean there are cases where contemporary artificial intelligence this these Brute Force methods have contributed to science so one of the hard problems of biology is the problem of what's called protein folding how do proteins the cells you know the DNA gives instructions for producing proteins and then they take a particular shape how did it happens well by studying you know huge numbers of cases artificial intelligence methods were able to give some insight that helped scientists discover how proteins are folding that's fine that's a very good use of it but uh maybe something will happen in these domains but how can you predict how could anybody predicted in the 19th century that 50 years later all of physics would have been overthrown totally reconstituted in totally new ways through relativity and quantum theory can't predict that and we can't predict what the next Century will bring maybe some uses of computational methods in the study of social systems it's not impossible [Music] um thank you thank you very much we run out of time and we are deeply thankful and grateful to your answers today that you shared your wisdom with us and I know there might be various questions from the audience but it was not possible I'm really thankful to our audience and particularly the panel that they prepared the questions for you and above all I would like to thank you for giving us this privilege to be with us here in Athens Greece and meet you in person and understand better your thought not only about your philosophy of language but also your philosophy of life and your understanding of life thank you very much we are really grateful to you thank you professions pleased to be with you thank you very much thank you all thank you all for your patience thank you very much thank you thank you [Applause] amazing

Channel: Institute of Philosophy & Technology

Views: 28,257

Keywords: Noam Chomsky, Artificial Intelligene, Language, Cognition, Philosophy and Technology, Institute of Philosophy and Technology

Id: _7AE7UuOfg0

Length: 65min 52sec (3952 seconds)

Published: Wed Jan 18 2023