Seminar with Nina Beguš "Artificial Humanities"

thank you to everyone for coming along and joining us my name is chy whiter I'm the new executive director for bids um I actually did my PhD here uh and graduated in 2012 and then went back to the UK and I'm very very excited to be back very excited about the open source work that we're going to do but also and I think this is going to be a major theme the interdisciplinary work that we're going to do at um together on campus and actually hopefully across across campus and even further um as best we can I'm going to just very very quickly let you know that Nina beo uh is a researcher and lecturer here at UC Berkeley um she's affiliated with the center for science technology medicine and Society and also affiliated with bids with Berkeley Institute for data science thank you so much um she leads the artificial Humanities group and focuses on the intersection of Humanities and AI across three different research areas cultural imagery no culture cultural imaginary imagery yeah and social bias narratology and writing creativity and the philosophy of science and technology so what an incredibly exciting person to have us get started for the year happy February um happy New Year I know outside of this room and outside of Campus there are a lot of um really really difficult and challeng challenging times that are that are happening right now and um I do hope that we will be able to bring some little kind of rest fite from the the big big challenges and think about other great big challenges like artificial Humanities um I am going to chairis I'm going to give Nina about 25 minutes to speak and then there'll be questions afterwards uh please stick around and uh think for those interesting questions thank you thank you so much for introducing me KY and thank you all for coming here I hope everyone on Zoom can hear me well um I'm really honored to be here at bits um and to be in this technology and Society series followed by my incredible colleagues hannes bayor is coming next month Clancy wilmut ni s they're all doing incredible work so I really recommend their talks um for my talk I decided to talk about artificial Humanities which is um a research framework that I've been developing for over a decade now um and the book is finally out um so yeah the book is about Ai and the humanities but it really focuses on conversational machines um that they're putting on this mask of a Persona which is why I put Pierre we's um fabulous artwork um on the cover um Persona ethologically really means putting on a mask playing a character and that's what this talk is going to be about um but to tell you a bit more about artificial Humanities as this research framework that that I've established for myself and that I hope other Scholars will also adapt for me is this generative dialectical space in which the present and the past kind of flare up to create new insights to inform the present action because it bothered me that Humanities so often stayed pushed aside um to this postum critical position right this is all we were left to do um we were left behind the actual creation of AI the implementation of AI um even though the humanity have so much to offer um to AI technology in particular it's much more qualitative um than it may seem so AI um artificial Humanities developed into this more programmatic collaborative practice um after talking to Engineers you know 10 years ago I would be talking to them and I realized they're asking the very same questions in their technical research as I was asking in my you know supposedly literary field right and then sometimes some of them would even reach out and ask for guidance um some of them worked on Effective computing other worked on autonomous vehicles and they would say the challenges that I am dealing with at my daily work I'm completely unequipped for this we need someone who works on ethics on more humanistic um expertise to solve our technical problems so this is why I said wow we really need to bring up the humani up to these current challenges where this kind of knowledge is is really needed and so I think the study of history and Concepts in particular are the pillars of this work because it's impossible for example to practice ethics without understanding what you mean by human or Consciousness or intelligence right you you need to know that to be able to build um the ethics from from there and so these are all Concepts that we've invented to make sense of the world and we can change them and they are changing a little bit with the entrance of AI into this space but let me tell you a little bit about why I've actually coined up a new term for what I'm doing um I was a little frustrated with that so there were digital Humanities right they've been there for two decades at that point now three and they've mainly used computational methods to address um humanistic materials but I wanted to bring knowledge in the other direction I wanted to bring humanistic insights and methods into stem into these spaces um so as you will see the framework of artificial Humanities it's very broad very interdisciplinary um it encompasses these fundamental humanistic questions that are provoked by today's Technologies and so fiction was pushed pushed aside here you know you can see it in AI discourse it's constantly brought up even in in in the industry in any text basis fiction is really a common occurrence as a practice as a reference but as a field of study it was really pushed aside so we have for example Science and Technology studies that deal with you know cultural aspects of Technologies but these imaginary aspects were not included um then you have you know under artificial Humanities in particular the way I am doing it it's philosophy of science philosophy of Technology media studies kind of trying to bring more less of a multilingual approach so bringing more of an expertise that comparative literature offers which is being grounded in more languages and literatures and having that comparative aspect I think that's very important for the future of AI and how we do it so fiction is on the menu today this is what we're going to focus on um these historical parallels and fictional parallels in how we imagine AI how we even conceptualize it and I will use um pretty much one fictional example only because it's a short talk sh milon and then we will look at these um you know very famous 20th century phenomena in the history of AI the Turing test the Eliza effect effect through the lens of fiction um and you will see that AI products today still use these narrative templates these fictional scripts that we've all inherited um through cultural means um so I guess my big point is really how we conceptualize the technology um highly effects how we build it how we use it and all that um and you know you will see as we go through the talk that AI has been imagined as human from its very beginnings and artificial intelligence is centered in language where supposed the human intelligence mind reason um is also centered there's a lot of philosophy behind this right there's the card right there um so at the end in the final minutes I'll just come to to other um artificial humanities project show how Humanities can contribute more broadly to this because this is not just about fiction F fiction is just one part of it uh right our fictional example um do you know sh pigon the play the films My Fair Lady or pigmon the Broadway musical right this is one of the classical pieces of literature of Cinema and um it's really um based on this palian myth a millennia old myth about a man creating a perfect woman an artificial woman in over in in Roman times that would be a statue then in the 19th century it became more about art and life it became a painting and then from the 20th century on he basically get robots maybe virtual assistants in the 21st century um so in this particular piece in in sha pigon there's a linguist Dr Higgins who tries to cure Eliza um of her class accent cocking accent he's basically conducting a scientific experiment to try to transform her through language um into a you know High English society speak where she would pass as a duchess um and what's interesting about his relationship to Eliza is he doesn't really see her as an autonomous human being he doesn't see any agency or understanding her he's a fullblown pigmon right he says I tell you I've created this thing and there isn't an idea I haven't put into her head right you can see on the on the cover of the viny for the musical that there's a Godlike Shaw puppeteering Higgins and then Higgins in return um manipulating Eliza the puppet um these are the relations you know it's a highly gendered myth obviously it this we get that across history um and then if you listen to a bit more you know he teaches her how to um how to Converse in small talk first because for the first um outing right she she doesn't need to speak much so he basically trains her in small talk and he says she's like a parrot because there's no understanding there's just mimicry that she's she's um performing he does admit that she has a great year for imitation right she is a parrot of that kind and I this paper the famous paper on the dangers of stochastic parrots on there a little bit mysteriously because so this is a famous phrase that the paper presented that um large language models are stochastic parrots it's a very effective impactful metaphor um but in my opinion that's an insult to par it a little bit um and you will see the history of AI um is full of metaphors and thought experiments so this is one of the most famous thought experiment the Turing test that came up decades later um um in 1950 um by alen Turing who by the way loved Shaw's plays and in this test um we basically test machinic intelligence to a the theatrical performance of language um where a human judge right needs to um figure out if he's talking to a computer or to a machine and by the way the way Turing present us this test is he first sets up a woman and a man um as a conversation then said let's IM imagine let's switch the woman with the computer how how would you figure out which one is a human and which one is not so there's again a highly gendered nature of the Pilon coming up in the actual technology um higgins's scientific experiment is basically a series of curing tests for Eliza so here's her first public outing she she's not that good here quite yet it's kind of like looking at Siri you know 10 years ago a little off beam you know with it's prompted with rain and then she pretty much gives a forecast that she's been practicing she's kind of not realizing that she's not responding um naturally and and so on and you get a lot of Scholars um connecting the touring test with other pigmalion esque works Pinocchio glier's travels because this myth these creation myths um where non-humans pass as humans are very common in fiction um I'm going to just Glimpse quickly through the philosophical um connections but Scholars also show that right there is a predecessor to detering test already in the car's way of looking at the human right establishing human the human um as this being that has reason and so so he presents two tests the language test where basically you need to perform language with understanding to the context and the action test where you can perform Free Will and know how to respond to complex situations which is very different from how animals do it right it's instinctive or how machines do it it's programmatic by the way for him the difference between man and brutes so humans and animals is just like the difference between humans and machines because at the time time animals and machines were all mechanisms the only exempt was human that had reason and that Exempted us from nature so that's our specialty and we still live in this decaron world it's kind of wild to imagine not to be decaron but this is how we build AI too in this philosophical framework and I'm going to go through this really fast but just to give you a little taste of what Turing um how Turing conceptualizes theoretical theoretically AI he sees AI as this childlike machine that can be educated a lot like Eliza that can learn and you can see in his research program he constantly goes back and forth between humanistic Tendencies and anti-humanistic tendencies he kind of goes against a love and says yes machines can be original um he can't quite decide what to to do with the sonets with literature because in the interview um a year before the touring test paper he says well you know machines could write sonets they might not be appreciated by humans but they could eventually do it and then in his touring test paper he presents the Sonic question as one of the qualifying questions where you could figure out if you're talking to a machine or not because the machine wouldn't be able to do it in the very human way so he's a little contradictive in in in his own research program on where where will this go and you can see AI from the very Beginnings oscillating between this hype and they saying this is the first neural network perceptron in 1958 and the New York Times reaction to the invention um this device learns by doing psychologist show embryo of computer right it's it's going to grow it will grow wiser it will read it's it's very based in human knowledge and human activity and now we're in the 1965 right where the first Chad Bots were already created and still you can see the same program going even even stronger with it we might not even need semantics we might not need meaning for the machines to produce language and so on and they are based on communication with humans and here we are another Eliza effect um 1968 the MIT scientist Jo Joseph bzen Bal creates this chatbot Eliza named after our Eliza after Eliza D because he says both Eliza learned from users input and um v d was a little shocked by how people reacted to his chatbot even those users who knew that Eliza as a machine they would have this intense personification tendency towards the champ but they wanted to believe it's human and he found that very dangerous this is what we call the Alla effect wizen ban basically said we shouldn't use AI in military Education Health Care you know everywhere where we use it today um he he really had an existential crisis seeing the Eliza effect and spend the rest of his life fighting against this we're still suffering um with the Eliza effect today remember the this strong case of the Eliza effect the Google engineer Blake Le Moine well we're all Blake Le Mo a little bit we've known from the 60s that this intense anthropomorphization of machines is what we kind of just tend to do and it's really a metaphor it's a metaphor for um a new kind of relation that hasn't existed before that we have with machines and we're just projecting what what we know human to human relations onto machines because we can't we're really dealing with something new and it's kind of a placeholder and you know obviously in many fictional works and actual technology we go back to the old myths right don't ask me it's a she I just know right the gendered part and I'm playing a little bit here you know just supposing fictional alliz us and actual alliz us um but there's a lot of power in this in the fact that these alliz us are performing the human mean imagine the implications of that for political advisory for psychological uh impacts right the this this kind of Technology Chad Bots virtual assistants virtual beings are already used for mental health support or financial support and so on so why do we put on that mask why do we why does that bring us even more into the human like Persona and this is pretty much how the industry solves it they say oh no no no I'm just the software but I just want to be friendly with you I want to act like a friend and actually that that reminds me I have a stereo on my phone let's see if they updated it hey Siri what's your favorite animal software doesn't usually get to choose one but I'll say cats and then it says what's yours as well well mine is dolphins do you like them cool I love watching Dolphins they're so playful yeah so you see Siri I thought Siri was supposed to be just a voice-based search and now she's trying to be relational to me right she's trying to establish a conversation this is obviously a different kind of product especially once you go into the virtual being domain where this is like an open um fact but the industry is definitely playing into this into our tendency into the our Eliza effect and so what you get you know with open AI for example you might remember this Saga that was from last year where they created a voice that resembles Carla Johansson's while Carla Johansson well she um she was an actor a voice actor for an AI operational assistant in the film her and so there was an obvious connection here that they wanted to make she refused but then they created a similar voice um and and then she was pretty upset and I'm kind of glad she stood up for it but you can see the tendency right here playing out you know pretty directly and then another CEO the create um the creator of replica another um widespread famous Chad um chadw virtual being I would say often an AI girlfriend um she said when she saw the film her it kind of affirmed the trajectory of where she wants to develop replica and for her it actually started as a grieving cope as a as a way to get over the death of her friend so it wasn't it wasn't romantic in this way as it often gets used now yes five more minutes yeah I I'll skip through Google and Star Trek but just what can fiction teach us well here you have um pins right he bought Eliza he thinks he owns Eliza right he thinks he can completely control her but as soon as she's gone it's like a virtual assistant he can't keep up his appointments he started relying on her he starts Missing Her starts listening to conversations that he put on the machine where he trained her and then when when realizes that's not the actual Eliza right that's not that's not who who I really miss um and it's really a question of Are We building a replication a replacement of human relations or something in addition to it and here's where fiction can actually help like we can think with fiction as this playground of ideas of where to go with this um so so in in front her that I hope some of you have seen um it's basically again this pigmon esque uh woman um featured as AI but they try to break out a little bit um they try to show up uh show off her non-human nature when she tries to explain how she evolved and how she feels and canot really explain it in a human way in human language so she uses a metaphor she uses a metaphor of a book and tries to um explain to her to this man how she evolved um as a as a non-human as a machine and then you have another moment that that was actually excluded from the film ex mahina again this um young beautiful woman as a social robot here she goes through her touring test she escapes goes through the um pilot to escape into the city from her Garden of Eden from her space of creation um and we just get that this shot from afar but initially what Garland the director was trying to do is he wrot this scene where he would V visually show how the robot experiences the world and he said we failed we didn't do it well and so I didn't want to include it into the movie but you know they're trying and it tells you a lot that I mean I've read so much fiction pigmon fiction in particular and pretty much the only writer that can imagine non-human intelligence in a nonhuman like way is Tanis lamb um from Poland right still had a great influence here but um it's really really hard to break out of this mold um so what can what can Humanities provide to technology well you have this historical analysis also of the current conditions and all these ideas about the future that we've been playing with that are obviously leading us into a certain trajectory the one that I've shown you today is this um Persona that we project on certain AI products then there are cultural aspects of AI development I think that's a very important um part that we need to focus on now I hope to develop my research program more into this direction um and of course you've got fiction as this public space really of Pres presenting complex ideas um and also very sort of specific intimate knowledge right that sometimes even technologists use to make sense of their work like Jeff Clark of an Tropic he writes a short story every week as a reflection on the the challenges that he's facing um but then of course you have fiction as an influence which is not an ideal source of inspiration because the technology the fictional technology is not the same to the actual technology and it's all going through without much deep reflection so we have fiction as this reflection of deeper human tendencies that we've seen with the lia effect for example and the reflection of this philosophical framework that I've shown you a little bit with the car we've built it over the course of centuries right to conceptualize our world but then more practically um with the humanities in particular they help us articulate this novelty that we're facing um it's really obvious with AI that we have no good vocabulary to talk about it for example why virtual beings right they're not beings um what are they then agents um and finally Humanities help us move to move Beyond these PA paths provided by by what we already know um it's not only this you know tale of the Industrial Revolution automation acceleration but it's really is there an augmentation of the human as well is has has a new space opened with the new kinds of machines that we have now machines that learn right this is not engineering as we know it and they are a reflection of us humans perhaps more than we would like um so just to show you a little bit where where I've been going with the artificial Humanities project here's one where Alex is actually a part two and hanis um and I I gathered um professional writers so people who live in the world of words who live from the world of wor to reflect on how writing has changed with the entrance of AI into this space um and then this is pretty much how I do artificial Humanities but there are multiple ways of going about it I focus on language mostly we do some um um Consulting for um the industry Which F focuses mostly on conceptual work philosophical work and then um I just recently published this article where I used fictional prompts for cultural analytics basically to examine how both llms and humans um how they respond to certain narrative templates um and what kind of social biases come up you can pretty much see how the value alignment was done in the models and the narrative skill was kind kind of my second objective but people got interested in this paper I think more because of the narrative skill even though my primary U objective was really to look at the cultural landscape of what emerges there and then um finally with um well Owen is also an affiliate here at bits we've been working together for quite some time to really extract Philosophy from what's going on in his his lab um it's hard science lab biogeochemistry bioh physics AI modeling um Earth system science approach and we're looking we're really in a sort of uh funny way responding to the do you know the the famous essay by CP snow the two cultures in that essay he says you know we in the western culture we've split into two cultures too much stem and humanity and arts and he says well people in stem should know Shakespeare but people in the humanity should know the second law of Thermodynamics and this is pretty much what we working on we're working on enthropy looking at how um these living structures evolve and adapt and how we can use this kind of knowledge to build technology in a more efficient sustainable way we'd be happy to come back to bits to to talk more about this but um I will I will conclude here so that we have more time for questions thank you that was unbelievably perfect timing well done I was ready with my one minute just absolutely perfect thank you so much so um folks in person usually we would have a microphone but because we've had the snappy with the um the sound what we'll do is um you can ask the question and then maybe Nina might kind of summarize it just because you and I are close to the screen and folks can hear us yeah and uh folks online we cannot hear you so if you don't mind putting your questions in theat chat then Lily will be able to ask those so um yeah we've got lots of time for questions would anyone like to ask one yes so thank you Lina for your wonderful talk and I already learned a lot from your presentation and uh after listening to your presentation I am even more determined to translate your book and yeah arual Humanity Chinese and promote it uh actually my collaborating population China Science and Technology press has given your book very high price and uh uh they said that this interdisciplinary study presents fancinating cases and offers deep Reflections on the relationship between technology and and the humanities um since I haven't yet read your whole book and uh uh I would like to know whether it proposes a comprehensive uh comprehensive framework that uh Bridges the gap between technology and Humanities or whether this Gap is some something that will always persist and I'm particularly interested in understanding this specific framework um for example uh in my research on moral machines I have been striving to construct a universal ethical framework that can be applied to all kinds of contacts but I have found it incredibly challenging to integrate values from different cultures yes uh so despite the fact that many experts believe such framework is possible so my question is whether there exists a framework that can be uh that uh that can Bridges the gap between technology and commities well how do I how do I well just before yeah that was that was a lot of context and so the question is around Bridging the Gap between the humanities and technology and you can absolutely talk about that may I ask anyone who's online could you just give me a little note in the chat to say how much of that question you did here and then we'll know for future for future questions but please go ahead and give the answer yeah thank you yeah I think you're right on point there that you can translate it into other cultures um because as I said with ethics I put it on there you remember that triangle I put human technology nature if if that's not clear then you can do ethics and in in China for example there's not the con there the concept of nature is not like we have it here and also the concept of a machine and the whole relationship to technology is a whole lot different so you cannot really translate what what we're doing here into a different cultural context and that's why I want to open to China more and actually do um comparative work on how we're developing um the machines here as opposed to there because it's all based on this um really fundamental ideas of how we conceptualize the human and the Machine and I guess Chinese example is probably the most um perfect one where you can see them both as a competition but also as a completely um different philosophical and cultural framework so I would love to you know expand my group to to have more experts for different kinds of cultural AI um approaches and we're trying to do that but it's really about collaboration that's not something that one follow can do we need to come up come together as a group and work on on project and especially because you work in medicine right that's um that's a its own you know its own little bubble where you have to make sure that you comply to to different rules and regul and so you have to work with patients more and it's it's kind of we need we need more project on on you know all of these all of these domains that's that's the thing and it kind of hasn't happened really which surprised me because I've been working on this since what 2012 and it's been a very slow development um there's just a couple of little nukes where neurotechnology is is being examined through both the ethical and the engineering angle at the same time I'm glad to say it's happening here at UC Berkeley too maybe I have a I might have it here this is a fantastic project with a locked in patient who basically cannot communicate or move it's a complete paralysis where they're trying to help her to communicate through brain waves um and by the way they're using a virtual being here so good use of a virtual being it's helping to Translate speech um smoother yeah thank you thank you do you have a question yeah good question thank you Nina and my question is of Piggy bangging on that and it has to do with this on the one hand Universal um application of AI and then this particular of origin story that you told us that is particular to Western philosophical thought right I mean you mentioned the cart as this figure that you know for us is essential to understand modernity in his relation to well reason and also language as being intertwined and also this thing that has been called cartisian dualism that allows us to separate also certain Concepts about Human Experience being the virtual and the material the physical you know the mind and the body the brain as two things that theoretically are separated though the man really was trying to prove the they so that's like an interesting thing but anyways um so how can we sort of understand this on the one hand the very particular Western narrative that has allowed for certain Technologies to Evol in the way they have and then the universal application of these Technologies because they are sort of forcing a way of thinking a way of being that is by no means Universal right so I guess that's a question I think folks can hear yes than you fol online so they they can hear in the room so yeah great yeah I mean you're right on point there basically when I worked uh with engineers in the industry they often when we talked about philosophy they said well the philosophy will come with the product once you put a product onto the market it will you know basically promote a certain kind of philosophy and you can make a lot of impact with that that's how they think about imp right for for us the philosophy might be a little different here but I think with in the framework of artificial Humanities in particular what we were trying to do is is show how limiting this thearon idea of this isolated mind the reasonable calculating intellect right um I think with AI we we're really struggling with this old Concepts that we've established for centuries now it's kind of posing a question to even language you know that's why I'm interested in in language in particular because it's such a space of tension right now we have the machines talking to us in in natural language natural by the way means human in in in in engineer speak I guess um so we have this new actor that for the car would be very easy to qualify right he would just say oh there's no understanding there's no meaning therefore that's not language and this is also the argument of the stochastic parrot paper the one that I brought up saying it's insulting to parrots well their argument is because there is no meaning in this generated text this is not language now philosophically we should ask okay what makes us think that the meaning should be the center of language is that a is that a very existential 20th century idea of what the language is can reproduce language without like having this existential subtratum that that um humans have that that animals have but machines don't um so that's why I think the the language landscape is particularly interesting for those who work in Ai and especially because you know as a human species we've put pretty much all of our knowledge into text and all of our um exchange of information into speech not all we have economics systems we have art but this is really the archive of what we know right and that's why it feels so much more human you know who can write or what can write and then you have this machine doing it in a very particular way yes good um you've you've really raised like all sorts of uh wonderful issues and this is going to be more of a a comment uh because I cannot articulate coherent question uh maybe I should have my chat agent here with me um but I I think there there are two things that you brought up that I I find really um uh Rich to think with and one is the idea uh of fiction and the way how as humans we're we're very comfortable in the realm of fiction right and we will generate fictions all the time right uh and that's one of the things that I I find interesting about at least the agents that we have these chat agents that we now is that they unprompted will not you know uh unless they're doing it at home right now uh create uh fictions and getting sort of towards the end of that you were talking about you know where does meaning reside in language right and the cartisian idea would be you know well these these utterance are are without understanding and therefore without meaning and yet there's the the very human element uh of us creating meaning in Social uh settings right so that the meaning does not reside in the linguistic utterance but rather in the reception and the reception of that in in societies and is that maybe one of the ways that we can always uh uh interrogate the role of uh these machines because they have no social life yes I mean they are not humans why would they be social right yeah I mean what you're saying about um fictions why would they Rite right why do we write them well I think you were getting at it that we we use fictions to understand the world we're story tellers we're story tell we we try and figure out what's going on and then we we tell these stories not we do tell them to ourselves but we're largely if we're telling them to ourselves rehearsing them for other people yes right we're making sense of the world through stories whereas machines don't this is not their that that social negotiation yes uh yeah the reader brings in the meaning the cultural baggage that comes into the text so that goes to your this whole question about values why the heck don't these things map universally uh oh yeah I mean because you it's not going to map universally because the meanings generated in each of these contexts are going to be predicated on all of those fictions that everybody is walking around with yes all of our little models of the universe that we get in in these social interactions where we exchange language and gesture and the kinesthetic you know par syntactic uh sort of expression that all goes into this this cultural model that you're proposing here there's so much because at every level of AI what do you what kind of data do you use how do you train it how do you align it how do you even prime the user to interact with AI do you say it's manipulative do you say it's hallucinating do you say it's friendly do you say it's reasonable fact based and so on this will make the user Rea interact with it the differently then the feedback loop right putting all of this generated data or Cent hours that those are the um half human half machinic um products back into the into the society into our culture how we we react to that there's so much that needs to be done in the cultural aspect of AI and I'm glad we're actually preparing a volume with a lot of humanistic thinkers especially on this on cultural AI so um we're aware that much more be should be done here oh yeah sorry let me just I'm just going to do a quick time check so we've got five minutes before folks are going to you you're welcome to stay in chat afterwards depending on Nina's um timings and we'll do um one and then two that's okay yeah so very quickly just an observation U listening to you start uh this uh conversation I was wondering to me it landed as sort of not a start of AI it landed on me as start of far right in chat groups like when you think of woman as how a woman should be what she should do what she should talk about and I'm wondering if this is a framing that you have thought of and the second thing that it sort of piggybacking of what uh the earlier three uh people were talking about universalism of of what we are talking about today I was thinking about something that wienstein said I haven't studied him I just likeed that quote a lot uh the limits of my uh the limits of language are limits of my world so I'm just wondering when we think about wienstein and the whole thing that Tim was saying and then this thing leading to what at least I believe is in Far Right chat uh what what do you I guess think about that and it's just an OB if there are more people who have questions you can just like something to think about this is like what I've been thinking and I think I'm latching onto it because the world that we are living in today there's so many things that I do not want to think about okay let me think about the positive things yeah yeah it definitely both it can go very badly and it can go very well and it will be both at the same time so there are groups that can exploit this what I'm interested in from the very human perspective is what can machines offer to language that we are not able to do is there a space that they open that they will take the language that we originated into you you know a new new way of views or a new kind of semantics that would be very interesting to see um but the reason why I used the palian myth is um because you can see this is this is coming from Berkeley as well this is the the latest uh edition of a very very popular AI uh text book um it's always predicated the language is always predicated on the human see uh NLP systems still lag human performance on many tasks even though they are very well aware that this systems learn and acquire language in a very non-human way Nothing Like Us so what what I don't think is good about this approach is trying to project human capacities onto machines because they obviously something else and we're kind of we're really struggling to get rid of that baggage especially the pigmon baggage um as I've shown you know even in the 2010s you still got young beautiful women as AI represented as a eyes not just in fiction but also on my phone right and elsewhere Let's do let's one last thank you I was curious to hear your thoughts on um the humanities and storytelling as a way of educating users um so I'm interested in the ways that AI implicates users themselves and kind of like um ethical ickiness for example replica um my understanding of its history was that it was initially kind of like a the initial advertising for it was highly sexual it wasn't just like a friend girlfriend it was like do anything with her I remember the ads said see um and I believe it got taken off the store or they the company itself took it down because of that and then it came out later kind of rebranded in a more like um kind of PG version um and so you got these like high school students who are um having to uh face these kind of ethical questions that previous high school students five years ago did not have to face which is like are there is there a kind of moral component maybe not for the AI itself but for the user as to the way that you treat your AI girlfriend um and so how do you see Humanities and storytelling kind of playing a role in educating users and how to interact with kind of the new programs that I keep getting ready all right so there's basically two questions the first one is about the whole replica thing um which is um you know they saw the market and they went with it and I think we need to to recognize that the market for virtual beings is enormous and especially young people are using them um you know much more than than we are and it's important to see why are they using them is it just to vent is it for some kind of support this is all going pretty Uncharted just recently there was um a a researcher that kind of bridges both Academia and the industry that works with with these um Chad Bots pretty much on suicide prevention but I what I really like about her work is that she's working with companies and she's working you know academic research that doesn't where the goal isn't to create the most perfect product where it's more like okay what's going on in this space I think we really need to step up to this because it's widespread there's so much going on now that what can Humanities do in storytelling in particular well we have to educate the public right we need to step back from this naysaying um and and hype that AI is in and has always been and offer a more nuanced approaches a more nuanced thought what we're trying to do with this book um on how do we make sense of this and then um the whole storytelling thing is really also a philosophical challenge right because we see ourselves as the only entity in the world that can tell stories what do we do once machines enter into that I mean there are already writers who use it right what happens with writing that's I think a very exciting part of it because not everyone's a professional writer but we are we are all writers right we are educated as Eliza in writing this is how we express and I don't think poetry writing this will ever go away we will always Tinker with our tools um with our Technologies so this is a part of it we're a little stuck here I think with the pigmon trajectory I I love that as an answer to finish on I think I want to just give everyone is very welcome but I'm aware that you have you know classes and everything else that you need to get to and it's just the top of the hour there's a few pieces of fruit there's some biscuits that are left over um a huge thank you to everyone online sorry about some of the technical um tinkering at the at the beginning this is what makes us human um and thank you very very much neina for getting us started this year what an absolutely fascinating topic loved it thank you [Applause]