James Brusseau: Desirable is the uncertain.
[Unintelligible crosstalk]
Other speaker (context): Hi, everybody. Good evening, and welcome to tonight's Center for Ethics event. My name is Ioana Chatzidimitriou. I'm Associate Professor of French, and I'm half of the team that is bringing you this year's theme, Transform, Translate, Transcend. The other and better half of the team is sitting right there. Josh Brzezhevsky, Assistant Professor of English, and the Director of the Writing Program. We would both like to thank the Provost's Office for their continuing support of our programming, our wonderful administrative support staff, Other speaker (context): Lori Dimmick and Karen Norman, who are not here tonight, but nothing would work without them, Other speaker (context): And the Muhlenberg Office of Communications and Marketing for helping us publicize our event.
Other speaker (context): Tonight, we are joined by Dr. James Brusseau, a philosopher, and professor in the Department of Philosophy and Religious Studies at Pace University in New York City, where he teaches and publishes in the history of philosophy, continental thought, Other speaker (context): And ethics. Dr. Brusseau has taught philosophy around the world, from Mexico City to California, Other speaker (context): Now in New York, and he also serves as a visiting professor in information engineering and computer science. Did I get that right? At the University of Toronto in Italy. He holds a PhD in philosophy from Pennsylvania State University.
Other speaker (context): Over the course of his academic career, James Brusseau has written across a wide range of topics. His earlier books include titles such as Isolated Experiences, Gilles Deleuze and the Solitudes of Reverse Platonism, Decadence of the French Nietzsche, and Dignity, Pleasures, Other speaker (context): Vulgarity, Philosophy, and Animal Rights. He is also the author of widely used textbooks such as The Ethics Workshop and Business Ethics.
Other speaker (context): In recent years, Dr. Brusseau has turned his philosophical attention to artificial intelligence and the ethical issues raised by data prediction and creative algorithms. Many of his recent essays and papers examine the human consequences of recommendation systems and predictive analytics that acknowledges that curate what we see, hear, and read online based on our past behaviors, part of what he will be talking about tonight. In this work, he draws attention to a tension that is familiar, even if we haven't named it. These systems can create a kind of technological euphoria by perfectly tailoring content to our preferences, but doing so can also pose challenges to freedom and identity when our past becomes the basis for who we are expected to become. In his work, Dr. Brusseau also challenges conventional approaches to AI ethics. Instead of seeing ethical ethics as the basis for who we are expected to become, he Other speaker (context): Proposes ethical reflection only as a break on innovation. He proposes what he calls acceleration ethics, a way of thinking that embeds ethical inquiry within the very development of technology, treating innovation and ethical reflection as partners rather than adversaries. Beyond the immediate dilemmas of algorithmic life, James Brusseau's work reaches into some of the deepest questions about human futures.
In essays on the philosophy of AI avant-garde, he maps three ways of thinking about AI. The first is to think about the future. The second is to think about how human identity might shift in relation to technology. Post-humanism, which reimagines what it means to be human in a radically transformed future. Trans-humanism, which focuses on enhancing human capacities. And gen-humanism, which is the first time I encountered that word, a notion he develops to capture the present-day ways in which generative technologies are already reshaping who we are. As we continue this semester's programming, Dr. Brusseau invites us to reflect on the central theme of transformation by asking, how do technologies shape experience? How do our identities interact with Other speaker (context): Systems that predict and produce content? And how might ethical thinking guide us as we live with and through these innovations?
It is with great pleasure that we welcome James Brusseau to share his insights with us tonight. Thank you.
James Brusseau: Thank you, Well, thank you for that generous introduction. As I explained before, I am a para-ambulant or para-parathetic philosopher. I wander around when I think unavoidably, so I'll be doing that tonight. Tonight, what I would like to talk about this evening are three things. First, I'm going to talk just a little bit about one definition of transformation. That's one of the things we're gathered here in this series to discuss, so I'll present one definition of that. Then in the second part, I will want to push that definition to a theoretical extreme, to try and get a kind of pure concept of transformation. And then in the third part, we'll take these philosophical conceptions, conceptual ideas, and we will apply them to contemporary artificial intelligence in the area of recommendation algorithms. So those are the three steps I'd like to take this evening.
And to begin with transformation, I will offer this as an initial thought. The transformation is something that happens. When, instead of you having interests, the interests have you. That is, it's not so much that you are in control of your situation, selecting one or another item, but rather it's because, or rather, it's because an item has been selected, it determines your future selections. So when you make a kind of choice, which is sufficiently extreme to change the horizon, of your possible choices, then I think something like that might be one definition of transformation. And as I say, the simplest way to say it is that in that moment, you don't have the interest, but the interest has you. It remakes you. And maybe in your lives, a sort of simple example of that might be something like changing your major. Sometimes you have an interest. Like, I have an interest in economics, so I'll take an economics class. I have an interest in sociology, I'll take a sociology class. But when you change your major to physics, that will change everything. Not just the classes that you need to take subsequently, but the way you look at and understand the world. So that might be a simple case where it's not that you have an interest in one or another class, but the interest has you and makes you. And so maybe we can start that as an initial definition of transformation.
Now, the woman you saw on the last slide took these journeys. Her name is Isabel Eberhardt. She is sort of a semi-famous cult hero from a minor aristocratic family in Geneva. She, in essence, ran away from home, dressed up as a sailor, stole her way onto a ship, escaped to North Africa to see what she could find. What she found was different religion, different diet, an entirely different way of living. And what is curious, or one of the things that's curious about her life is that when she took this trip, and she was kind of a writer too, when she took this trip, her first trip, it wasn't that she was running away from something, and it wasn't that she knew what she might find there in North Africa. Instead, it seems like she was running away from something. And it seems like what interested her was the fact that she did not know, right? So why does one travel somewhere? I mean, it seems obvious you might travel because you're trying to escape a bad family situation. Or maybe you're going because you think you're going to Paris because you think, there, I will find love. But there is this third option that you're going because you don't know what you will find, or what you will become. And if you read her work, I think that's why she went. And so I'm calling that a one-way ticket. And I think something like that might be, in essence, an example of pure transformation. It's when you go somewhere because you don't know what you will find. And that's the reason that you go.
So that then converts into an ethics. The ethics can be that what is good is the unknown. What is desirable is the uncertain. It's something we want, not something we don't want.
And later on, we're going to talk about technology and artificial intelligence. But I think that this is something that we're going to talk about. And I think that this attitude is extremely important in today's debates about technology and innovation. At dinner, we were talking about this tension between innovation and safety. That is, every time you have a technological advance, you have some safety preoccupations. So AI advances in healthcare have saved lives. But they've also spilled all kinds of private personal information onto the Internet. And it won't be long before people of your generation maybe will not be able to get health insurance because a company has discovered something in your personal information or your parents' personal information that makes you high risk. We don't know. We don't know what's going to happen. But the question many of us are facing today in artificial intelligence is, should we go ahead anyway? Some people say yes, some people say no. I am not going to take a stand on that. But I think that one reason why some people want to go ahead today is because of that ethics of the unknown. It's because they find what we don't know, what we're uncertain about, to be something that is worth having.
So maybe there's this strange connection between this 18-year-old aristocrat in Geneva, Isabel Eberhardt, and today's AI, artificial intelligence, pioneers. Maybe there's a strange closing of the loop, that they're both driven by that same ethics, that the unknown is good, that we want a one-way ticket.
Okay. So that's part one of part one. Part two of part one is, maybe we have reached the end of the one-way ticket, and that's because of social media like Facebook. Isabel Eberhardt and, giving away my age, me, when I was young, I could go. I did. Like many people my age, I went to Europe, I went to Thailand, that kind of stuff. And when I arrived at those places, I could completely create a new identity for myself. I could change my name, change what I was interested in, just explore a new life. I can't do that anymore. I couldn't just roll into Allentown tomorrow and try to start a new life. Because people would find my face. On Google, they would see my profile in LinkedIn. It's almost impossible, going back to the notion of transformation, to have the kind of pure transformation that Isabel Eberhardt discovered. One where she escaped, for reasons she didn't understand, to a place where she could have new diet, new interests, new religions. Now, for me, that's almost impossible. My old jobs, my old friends, all of the old things I used to have, they chase me around on social media.
So, I think that there's a problem in artificial intelligence today. The problem is, one problem is, it's very difficult for us to escape who we are. It's very difficult for us to have that experience that Isabel Eberhardt had of buying a one-way ticket. I mean, existentially, not geographically. Because of Mark Zuckerberg, and Facebook, and others like that.
Last point, conclusion of the first part. One argument that I'm going to want to make tonight is that while technology causes human problems, technology can also solve human problems. That is advances in or manipulations of technology can solve the same problems that technology itself created. So there's this man, Max Hawkins, who is kind of like the woman we saw in the first slide. You can look him up on Google. This is not AI. This is a real thing, a real person, real life. He did the following. He had this same sensation of being trapped. He was trapped by Facebook, being unable to enter uncertainty, to find interests which were unfamiliar to those that he already knew. He couldn't get out of who he already was. So he hacked Facebook, his feed, to recommend to him things at random. So for example, he was recommended to go to some Czech bar in some neighborhood that he never would have visited before. So he recovered from Facebook the very thing that Facebook had taken away. He found a way to twist the technology in order to help him find interests that were unfamiliar that he would not have discovered on his own.
And then I want to say it to conclude this first part. Maybe that's a way for the old idea of transformation to be reborn today within artificial intelligence, by hacking into some of these social media sites and twisting their algorithms against themselves. We're going to see a more elongated example of that in a bit. But that's the first part.
There is an idea of what transformation is. Transformation means developing interests which are so new that they open up a whole new world. Transformation means developing interests which are so new that they open up a whole new world. Transformation implies an ethics that the unknown is desirable in and of itself. And social media tends to resist that. It tends to disallow us from discovering things that are completely new to us. But maybe there's a way out, like Max Hawkins found. Okay, that's the human start to what I want to talk about.
Now I want to just push it a little deeper conceptually. These ideas. And we'll start with a quick thought experiment. This is a thought experiment I like to use to talk about human freedom, but it works equally well for transformation, I think. This is from John Locke. It's the 17th century. And what he asks is this question. He says, look, imagine that you are in a room, like the room I'm staying at Hoffman House here on campus. It's great. It's large. It's warm. Windows. I'm very happy there. So you're in this room doing the kinds of things that you want to do. And while you're doing that, someone locks the door behind you. You don't know that it's locked. But there is a question. And this is the question that John Locke asks. The question he asks is, are you free? If you're in a room doing what you want to do, are you free? If you're in a room doing what you want to do, are you free? And you can't leave.
Now the philosophical answer is, there are two levels of freedom. There's the higher level of freedom, which is the ability to choose the kinds of things you're going to choose. And there's a lower level of freedom, which is the freedom just to choose what I want right now, like a drink of water or a candy bar. And that lower level of freedom, here you're still perfectly free, even though you are trapped within the room. But on that higher level, you are not free, because you're not able to change the kinds of things that you want. That's freedom. Let's think about transformation. If you were in this situation, would you care that the door was locked? Would you be against this? Would you resist this? I'm not sure that I would. If you're getting everything you want, then why change? Isn't the reason that we have freedom in order to get the things we want? So if we have them, maybe we don't need freedom anymore. Maybe the same thing goes for transformation.
So what I want to ask in the second part is, I want to ask, is transformation something we actually want? And basically, I think that there are two, this is something of a simplification, but there are two approaches to this, or two ways to answer this question. One is no, one is yes. So on your left, Martin Heidegger, I think that in a certain sense, he says no. He says the reason we have philosophy, the reason we have human freedom and volition, is to find out who we are authentically. That is, there is someone that I, James Brusseau, am. And what I want to do is find out who that person is. And if I can find out who that is, I don't want to transform. I don't want to be someone new. I want to be who I am. And if I'm trapped in that self, like I'm trapped in a room, that's good. That's what we want. The other side, this is Gilles Deleuze.
Gilles Deleuze has the opposite position. He says, look, the problem, Heidegger, is that there is no authentic self. There is no true person who you could be. So if we don't have that, then what we can, the best way we can live, be vibrant, be vital, be alive, is precisely by embracing that reality and transforming constantly. So Deleuze is the philosopher of constant transformation. He's the philosopher of the one-way ticket. He's the philosopher who says, look, the unknown is intrinsically good. I want it. And maybe it will create a new me. And that's also good. Heidegger says, eh, not so much.
Now, how do these two look in the world of today's technology? Here's what I want to say. This experiment on the side, the hand. This is a good experiment. I like this one. This is run by the Cheetos company. And the Cheetos company asked their consumers, what do you like about our products? And what do you not like about our products? What all the consumers said is they hated the little dust, the gold dust that they got on their fingers, made a big mess. But otherwise, they loved it, loved the product. Okay, so the Cheetos executive said, sounds good, sounds right. But the Cheetos company said, well, let's just check that. Let's just check. And so they strapped one of these sort of electromagnetic casks, helmets onto these same consumers and asked them the same questions and watched how their brains lit up as they held the food, as they played with the food, as they ate the food and so on. And here's what they discovered. They discovered that what people really liked about Cheetos for some deep genetic reason, who knows? Was the little mess on their fingers. What they said they didn't want is what, in fact, they did want. That's what turned them on to the Cheetos. They just loved the feeling of being like a child and having this mess on their hands. So if you look at their campaigns today, Cheetos actually has a little person who intentionally throws this dust everywhere. Even though people say they don't want it and think they don't want it.
So there's this kind of convenience that Cheetos is giving us and that I'm calling euphoria convenience, which is one where you get what you want despite what you want to want or what you think you want. Now, this entire idea depends on there being some kind of truth about what you want, an authentic certainty. Maybe there is, maybe there isn't. I don't know. But if there is, then something like that, the hand is a Heideggerian position. There is something you truly want, and you want to get that, and you don't want to transform away from that.
Now, at my university in Italy, in Trento, one of the tools we're working on has to do, we're big into movie recommenders there. One of the tools puts people into one of those electromagnetic readers and shows them movies, and we try to determine from their brain. Reactions, what the perfect movies are for them, as opposed to asking them, do you like action-adventure, do you like comedy? Don't forget that. We say, let's just go to the deep human truth. Let's see what's happening in your brains as you're watching a movie. Now, if this were a philosophy class, we could talk for a long time about whether or not what you see in your brain really is the truth about who you are. I'm not sure. But for our argument today, let's assume it is. If it is, then the powers of technology perhaps have the ability to give us what we want, to show us who we should be, despite what we think we want, despite what we want to want. It could be that that technology could satisfy Heidegger's dream. Technology could show us who we are.
And we see this every day in predictive algorithms, in Netflix telling you what movie you want to watch. LinkedIn will show you one job or another job, but at the same time, they're taking away millions of jobs. They're saying, you are this person, or you are that person, professionally. Hinge does the same thing. This is who you are as a husband or a wife.
Now, I'm not sure how accurate those tools are, but this is a conceptual moment. And conceptually, it's possible to imagine that those tools know more about us than we know about ourselves, as it occurs maybe in this hand example. And if that's true, then we reach this state, as I call, euphoria convenience. That is, it's where the things we want can be predicted. So there's a black mirror scene where a couple sits down in a restaurant, and after sitting for a few minutes, the waiters just arrive with their food. Well, why? Because the restaurant predicted what they would want. Don't need to ask. That, I want to say conceptually, is something like pure authenticity, and it allows euphoria convenience. You get what you want before you even know you want it yourself.
On the other hand, there's Isabel Eberhardt and Max Hawkins, and what I'm calling freedom squared transformation. And this is everything we said in the first slide. The idea that you can not just choose one thing, not just choose, I want the water, I want the Coke, but you can choose to choose. You can get above that. You can say, I'm going to embrace something which is unknown. I'm going to change my major to physics and see what happens. I don't know if I make that change. I don't know what I'm choosing. I'm choosing to choose. That's freedom squared. That's radical freedom. And that's radical transformation or pure transformation. And that's what I want to talk about today. So second part, I think it's just about over.
The second part is this, going from the first part where I talked about the one-way ticket and some superficial definitions of transformation. Now, this is the pure concept stuff. There's two extremes. Pure authenticity. You are who you are. And perhaps AI can tell you that better than you know yourself. And you should become that person. I mean, if that's true, if there's someone who you are, you don't want transformation. Why would you? Just like if there's something you truly enjoy, you don't want freedom. You don't need freedom, as long as you're already getting it. What do you want freedom for? By contrast, if you're on the other side, if you're on the Deleuze side, then you say, look, there is no authentic self. And in that case, we want not just freedom to choose, but freedom to be free. We want to choose to choose. We want the unknown and see what happens next.
Okay, so to conclude Section 2, I think that there are these two really fundamental ways to see the world and to understand human identity. And I think they've become clearer today, 2026, than they were only 25 or 50 years ago. And they've become clearer because of technology. Technology has allowed us, it's been part of what has allowed us, to push these two conceptions of the self to these extremes. The Heideggerian idea of authenticity, which is one where it's not just that you're true to yourself, but you're true to yourself and you do not want transformation despite what you want or think you want. That's radical authenticity. And then there's, as I say, then there's the freedom squared, the idea that you're not choosing one thing or another, but you're choosing to choose. You're embracing the unknown itself. Okay, that's part two of what I wanted to talk about.
Now, the third and concluding part is, all right, that's great philosophy. I'm in the philosophy department. Sounds good. What does this have to do with the real world? So what I want to look at now is how these ideas connect with one project I've been working on in the real world.
I think I was explaining earlier before how different it is. I work in the philosophy department and in the computer science department. And in philosophy, that's my youth, we would just write papers and be happy. And in computer science, it used to be, when I started kind of about five years ago, people would write papers and then they would start businesses from the papers. But now it's gotten gone so far that people start a business first and then they write a paper. So this is one of the businesses that I've started. It's called the Curiosity Engine. And now the papers are just starting to appear from it.
And this business and the paper was created as a response to what I just located before. Okay. This idea that technology is hemming us into an authentic or singular version of who we are. And I wanted to ask how we can get out of that. And I wanted to ask in a specific way, how can I be like Max Hawkins? How can I use the technology which is trapping us to let us escape? And this is what we're doing with a small group in Italy. We are looking at recommendation algorithms. And when you look at it, when you look at recommendation algorithms, what everybody uses as the example is Netflix. Because it's just the example everyone understands.
So I want to show you how recommendation algorithms, recommenders work. And then how I'm proposing that we change the way they work. So here's how they work. What Netflix wants to do is they want to give you a movie that you're going to enjoy and want to watch. They have two strategies, basically. The first strategy is called content filtering. Content filtering. Because, I mean, AI, another point I was making earlier, artificial intelligence is not complicated. It requires a lot of compute, a lot of power, but the ideas behind it are not terribly complicated ideas. So here, this is an AI engine, which will choose movies for you. What does it do? If it's going to predict a movie that you will enjoy, it looks at two movies that you have already enjoyed and just filters for keywords. It's not any more complicated than that. The keywords here are Harrison Filter, Action Adventure, Drama, Fast, Risk, Police, Bad Guys, Good Guys, right? All those things are in common. There are some words that are different, like Future, Present, Space, Chicago, etc. But you just stack up the keywords. You find the keywords that work for a person, and you predict the next movie. Obviously, what's going to fit there? I mean, Star Wars would fit there, I guess. So that's content filtering.
You're just one movie to another. The one they use more frequently is collaborative filtering. Collaborative filtering is you try to find someone who likes the movies you like and then try to find another movie that they have liked that you haven't seen and then recommend that one. All right, so here's two people. They're similar. Similar as could be. And they both enjoyed The Fugitive and Groundhog Day. And the first person liked The Godfather. So what do you recommend? This guy, The Godfather.
Now, in both these two tools, the collaborative tool and the content tool, what's happening is you're using a logic of similarity. Things are the same. We'll come back to that. And it makes sense. It works. If I was starting out, that's the way I would do it.
So it's almost a cliche at this point, but it's worth repeating. We create the technology, and then the technology creates us. If the way that we are predicting the next movie or the next thing you're going to want to watch is by looking at what you've already watched or already enjoyed or what people like you have already watched or already enjoyed, what are you going to get in the future? You're going to get Instagram, right? You're going to get Instagram. You get into the little loop, and it's the same.
For instance, I've been getting in recently this robot demolition derby. There's like these little cars that go around and they try to destroy each other. I don't know how I got into that, but over and over and over again, I get these little robot demolition things coming in. Of course, that's the way the tool is meant to work. It draws you into what presumably you like. Now, of course, all of us are too complicated to be trapped in just one kind of real. But I still think conceptually, there's no difference. It's the same point we were making earlier about euphoria convenience. Why would I leave Instagram? I'm getting what I want. And it goes, I'm just as bad as anyone else. Like, what happened? Ooh, an hour. How did that go? Instagram works. And I'm happy that way. Maybe that's not a bad thing. But it's definitely a result of what I want to say. It's a result of the technology. The reason we're caught in Instagram is because of the way things are filtered. Because the things that are predicted that we're going to want are things that resemble what we have been in the past.
Now, we're coming toward the end of the presentation. So here's what I want to talk about now. How can we change the recommender's to get out? How can we switch them so they show us new things? I'm going to show you. There's two parts to this answer. The first part is not mine. The second part is mine. This part is not mine. This is just in the community. This is called orthogonal recommendations. And what you're doing here is you're doing just about the same thing that the standard tools do. You're taking, I mean, you collect here on the right where Her is, that movie. ChatGPT made this. So it's made with movies I don't know. But you collect here on the right. You like the movie like Her. That's what you like. And other movies like that. You dislike the ones on the left. And that's either you or your lookalikes, people who are just like yourself on the Netflix database. So there are all the movies that you're going to like over here, all the movies you're not going to like over there. But then there are these orthogonal movies. These ones that don't have any connection. The ones that aren't like the ones you like or you don't like. And I want to say that those are where we should look to try to discover new possibilities, to try to transform the way we talked about transforming at the start. Not by choosing some new little difference, but by choosing to choose. Freedom squared, rather than the hypothetical freedom. Get exposed to a kind of movie, let's say, which will open up an entire new way of thinking about the world or an entire new set of films that we might want to watch. So we find these. This is almost like the Max Hawking idea. We find ones that are unlike the things we've liked and unlike the things we don't like.
Now the tools do actually do this. This is called orthogonal recommendation. Netflix does do a little bit of this. But what it comes down to basically is this. This is basically just random. What Netflix is doing when they're trying to what's called explore for possible new interests, they're just throwing you random stuff, see what happens.
What I want to ask is whether it's possible to do better than random. Can I find a way to present to you new movies or someday new experiences which are unlike the things we've liked and unlike the things we've liked and unfamiliar, unlike those you've had in the past, unfamiliar to what you would imagine you think you want. But you somehow still find them engaging. They're not just random choices. They're something you might connect with. That's the heart of what I want to talk about today. And it's captured here.
This is the way the recommender that I'm proposing works. The score, the recommendation, the thing we're going to do, the number like 1 is 100% yes. We're going to recommend you that number. The score is that the cosine similarity, that's the first thing we saw. That's what you like, what you don't like, or orthogonal. The cosine similarity is going to max out or that function is going to max out when you get that orthogonal value. Second part, partial alignment. I'm going to come back to that. Third part, novelty. Of course, the movie has to be new. It doesn't count if I give you one that you've already seen. So you have to have one there. Quality. We want the movie to be good. Like if it's just a bad movie, well, who cares? It should be decent. And then coherence. Coherence means there are some movies which are just oddballs and don't fit anywhere. Those are too easy. Like, it's actually on one of the slides. What is that? Napoleon Dynamite. No one knows what to do with that. That's too easy, right? So coherence means the movie is like other movies. It belongs in a constellation of other movies. So those are the things we want to take into consideration.
But the partial alignment, what does this mean? What this means is I am wondering how I can find a movie that might be interesting to me, even though it has no connection to the movies I have liked or disliked in the past. How can I find one that might be connected with me anyway? And my answer is this. It's this idea of resonance, which we're going to talk about with Gilles Deleuze in a moment, but let me just explain the idea first.
So if you think like white wine and red wine, that's similarity. So someone likes white wine, and they're probably going to like red wine too. They're both liquids. They're both alcohol. They're both the same amount. They both have the same kind of elegance. They both have the same social role, more or less. So going from the white wine to the red wine, that's the way Netflix recommends movies, tries to find things that are similar. But there's also something strange. White wine goes with fish, but it doesn't go with meat. Somehow, despite the fact that they are different, liquid, salad, food, drink, etc., despite the fact that they are different, those two things go together, white wine and fish. But white wine doesn't go with red meat, and red wine doesn't go with fish. It does go with red meat. So there is this, what I'm calling resonance. There is this connection that can be established across things that are different, that are not similar. Even though they're unlike things you have seen before, you still resonate with them. They still connect. They still relate to you.
Where does this idea come from? This idea comes from Gilles Deleuze, Difference and Repetition, 1968. And if this were a philosophy class, then we would talk for a long time about the difference between the philosophy of identity and the philosophy of difference between Plato and Nietzsche, and whether the world is fundamentally one, that's Plato, or fundamentally many, that's Nietzsche and Deleuze. But this is not a philosophy class. So we won't go there. We will just say that one of the things that Deleuze talks about in Difference and Repetition is how you can, he uses the word repetition, how you can see the things that are fundamentally different repeat, but we'll say connect. How can things that are fundamentally different be connected? What he calls repetition, I'm calling it resonance. I think that's a little better word. And that's what I'm trying to locate here in the red wine and fish. And white wine example.
I'm trying to say, look, there's this resonance. There's this, we don't know what it is exactly, but it's there between the white wine and the fish and not between the white wine and the meat.
Okay. So how would that work for human beings? Here's what we're proposing. We're proposing something like this. We are going to change the Netflix strategy. Instead of starting with people who are the same, we're going to start with people who are different. Like this punk guy and this friendly suburban housewife or whatever, right? We're going to start with people who are fundamentally different. And we're going to try to recommend back and forth between those two people. Instead of finding people who are the same, find people who are different. But it just can't be anyone who's different. If it were just anyone who's different, then that's back to that random stuff again. What we want to find is people who are different, but in a way that resonates. Like white wine and fish. And maybe to do that, we look, and we were talking about multimodal AI earlier.
This is exactly this. This is taking advantage of multimodal AI. We're going to look at different ways that people relate to each other. Not just movies. Not just movies. Not just food. But all kinds of different ways. Actually, multimodal is like video plus voice and that kind of thing. But this is a different way of thinking of multimodal, right? We're going to look at different ways that people relate to each other.
So here in this example, we're saying that these two have different movie preferences, different ages, genders, nationalities, languages, et cetera. But for some reason, they share the same intense interest in a certain kind of restaurant, like Japanese restaurants, or even a smaller niche. So that's what we're looking for. We're saying, can we find people who are different in almost every way, but they share an interest in something really precise and niche, like a certain kind of jazz music, a certain kind of Japanese, not just Japanese, but like northern Japanese cuisine or something like that. And if they resonate in that way, then maybe they will resonate in movies. So we will recommend to her the kinds of movies that he likes and recommend to him. The kind of movies that she likes.
So it would be something like this. So he likes one kind of movie that's just completely different than what she likes. But because they have this resonance, maybe we should recommend one of his to her. Just like white wine goes with fish, but not red meat. And if we can do that, if we can do that, then we can make a difference. We can change the way AI works in the area of recommenders. The way it works now is against transformation. The way it works now is to always propose for you movies and experiences, partners on Hinge, jobs on LinkedIn, that resemble those you've had in your past. It traps you in your own past. What I want to say is that AI can help us get out and be transformative. Not because it's going to give us something that we want exactly, but because it's going to help us want to want. Choose to choose. It's going to help us discover a kind of movie that is different from what we would have imagined as possible. And that might open, new kinds of films or experiences that we wouldn't want to engage in the future. This is something like today's version of the one-way ticket. This is what Isabel Everett would do to discover a new reality today, as opposed to getting on a boat and going to North Africa. Okay.
So, that's the presentation. I think I've just reviewed the basic idea of what a proposal is. That for recommenders, instead of starting with people who are the same, so whose interests are just like yours, or starting with the interests that are those you have already had, you start with a different system. Start with someone who is completely different from what you are. Then find one similarity that you'd like to work with, see what kind of question they have. And then you recommend between them. And again, this is the purpose. The purpose is to open up, not to find some one single thing that's interesting, but to find a whole genre of things that are interesting. A different way
Other speaker (context): Of living.
James Brusseau: People went back to the start that's transformation according to this presentation okay so that's what we have for today on the subject of transformation and please feel free to ask your questions yes
Speaker uncertain: Interesting
Other speaker (context): Very good yeah presentation. Could you tell us a little bit more about that two-dimensional chart, like the movies on the x-axis and the y-axis, your movies there? No, just quickly, here's the business if you're interested.
Other speaker (context): Oh, yeah. So could you tell us what the x-axis and y-axis represent?
James Brusseau: It could be. It could be. There's at least two possible ways. One is, let's just say that the vertical arrow, that's me personally, and it's my, the orientation of my interests. And that orientation is determined by taking all of the films in the world, labeling them in terms of, let's say, a hundred, let's say, a hundred different things. And finding out where I point in that constellation of points. Then you hammer it down to two dimensions. There's the movie Her, that's a movie. You can imagine other movies, too, that are in the direction of things that I like. And Netflix knows I like them, because I watched them all the way through. I gave them five stars. Netflix knows what I don't like, also, because I stopped watching them after I went a little bit. So Netflix knows the vector for the things I like, and Netflix knows the vector for the things I don't like. And they know which movies, they know I don't like Her, they know which movies resemble Her. Like, according to either the content, the keywords, or the movie. Or collaboratively. So that's where this, that's what this is representing. And as I say, what Netflix also has, they have a few ones. Most of the ones, they don't have sufficient information. It doesn't fit into what I like, or what I don't like. Or what people like or don't like.
Other speaker (context): Yeah, then the numbers in the x-axis, like from the negative one to one, will be the, your, some kind of, you know, scores. Other speaker (context): That's how much you like your movie, right? Yeah. Yeah. Then the y-axis number, also, some scores for you like it, or...
James Brusseau: There shouldn't be scores in there.
Other speaker (context): Ah, okay, gotcha, yeah.
[Overlapping speech]
Other speaker (context): Yeah, because that chart reminds me of the, Other speaker (context): Our, we call it representation space, or something, in AI, Other speaker (context): When developing the AI model. Other speaker (context): So, but we use not only the two dimensions, Other speaker (context): But we use the very high dimensions, like ten dimensions, Other speaker (context): Or a hundred dimensions, or something.
Other speaker (context): Yeah, so, even if your movies are not, if your movies are orthogonal, Other speaker (context): I mean, they are, you don't ever think about the movies, Other speaker (context): But in the other dimensions, they could be aligned. Other speaker (context): So, I think that resonates that, you know, idea in that.
Other speaker (context): Right, right, but that's what gives you, when you lose something, Other speaker (context): We push it down to two dimensions, of course.
Other speaker (context): Yes, so that was very interesting, Other speaker (context): Because that resonates the AI training.
Other speaker (context): Yeah.
Other speaker (context): Because, yeah.
Speaker uncertain: Even
Other speaker (context): Even if it doesn't look like a fit, Other speaker (context): But in different dimensions. Other speaker (context): In different dimensions. Other speaker (context): Yeah, then it could be aligned. So, you could find something that may interest you, Other speaker (context): Even if it doesn't look like in this dimension. If it doesn't fit you. Yeah.
James Brusseau: Yeah, I want to be a little careful there. Because orthogonal is going to have to be things that you, that are unfamiliar to you. Things that just don't connect with those that you like or don't like. So, there can't be, I'll go back to what I said earlier. ChatGPT made a mistake by putting that, in the vertical axis. That shouldn't be there. It's just, it's unfamiliar. There is no just, which is not part of the continuum of what you like or don't like. But you're right. Of course, in the real world, it's not that easy. In the real world, where you have 10,000 dimensions, there is no movement which is purely orthogonal. Of course not. Of course not. But you can round it off with that.
Other speaker (context): With the concept of resonance, because there's so much more information taken into context, like, for example, I think the example up there was, like, Other speaker (context): You're different when it comes to demographics and different when it comes to even, like, the shows and movies, Other speaker (context): But then really similar in the kind of restaurant these two people like. Like, what are the implications of that for, like, privacy? Like, seeing as we already have very limited privacy on the internet, Other speaker (context): Like, if AI, like, let's say the Netflix example, if Netflix now also has the information about, like, what kind of food I like to eat, like, what implications does that have?
James Brusseau: Right. So, it's a huge problem because in order to make this work, practically, we need to have a lot of information about users. And the way it would work is just with massive information, because we would have to find, over the course of many, many, many years, where those resonances might exist. It's impossible to remember. So that's a real trade-off that Netflix is going to have to have. That's why I use the term multimodal. Multimodal doesn't actually mean that. Multimodal means that video plus language plus audio, that kind of stuff. But this is multimodal in terms of human experience, restaurants, jobs, and that kind of stuff. So how does that get solved? I would say look, that's a technology problem to solve. Making this work is going to open up a lot of privacy protocols. But that's a problem we need to solve. That's not a reason to stop trying to develop this. But you're right. Almost everything that works in the ER, almost everything, except in health care, almost in everything else, is going to work because it knows a lot about it. So we're going to have to manage that.
Other speaker (context): So I'm curious about your business. And I don't know how far along your business is in development. Okay, yeah. So, okay, what I'm curious was, like, the find the one solution. Other speaker (context): Similarity in order to get resonance. Do you have any data about what similarities are more efficacious than others? Yeah, that's what we would want you to find out.
James Brusseau: That's where the power of what people call artificial intelligence. It's just statistics. A lot of that's exactly where the problem would come in. That's exactly what we're filtering for. What are the connections that tend to be indicative of other connections? Does a shared interest in restaurants indicate other connections? Kind of latent interest? Or is it instead music? So that's a terrific question. And I can't answer it, but I can't answer it because it's the right question. And it's exactly what we are gathering data on. Exactly what we're trying to figure out. What are the resonant things?
Other speaker (context): Thank you. I have a couple questions, so I have to choose which one. Other speaker (context): Okay. What do you think are some of the implications for this as far as, like, political polarization? Like, the algorithms that we receive or sort of receive only information from a particular political perspective or community silos or, yeah, like a sort of perspective silo. So do you see any sort of implications that this solution could have for that?
James Brusseau: I do. We talked about that. It's been a few months in the past. Now, there's a group called PostSocial, which is interested in exactly this project and how technology can be used to help break people out of their filter bubbles, particularly with respect to news items and things like that. So there's a lot of research going on in this area in this respect to politics. The kinds of solutions they talk about, though, because they're not philosophers like you and I, so their solutions are less relevant. They use something called... bridging algorithms. So what they want to do, generally, is to try to break elected people who are 100% a favor of Trump, to break the muscle movement. 100% of the text people break the muscle movement. What they're trying to do to break people is they're trying to find news sources of information which are halfway to an opposing view. So I guess I'm not sure exactly what that would be. I'm not a big news person. But, like, let's say if you're really anti-a certain person, they might want you to read a I'll bigger person. Let's say you are anti-Trump, let's say, and your content is filtered all the way. Everything you see is Trump. You just can't escape it. It's a highly difficult box there. If you just show them the box, news, that won't help. It won't just keep them. You need to find what they're interested in, PostSocial. PostSocial.com. There are ways to take, like, little steps to get people out. So they're trying to find a news source for their feed, which is just a little bit to the right of New York, or MSNBC, whatever. So yes, the same research, that's a philosopher's answer. The better answer is the short answer is yes, the same research is happening in exactly that field. It's just the strategies they use are much more incremental strategies. And that, of course, is what we're trying to do here.
Other speaker (context): Thank you very much. It's very interesting. What struck me really about your whole talk and your project is that you framed the entire question of freedom, or the question of authenticity, on desire. Right? On the thing that we want, or we think we want, and we want, and we want. Right. But that seems to me to be a problematic or limiting way to think about what freedom is, right? If we reduce it simply to just what we want. So I'm not even exactly sure what my question is, but I guess it's something like, if you're thinking about this kind of radical transformation, are there other ways to think about transformation or about freedom independent of what we desire? But rather, I don't know. So in other words, just giving people things that you think they will like. I don't see how that increases freedom, really. It would seem to me like opening up all possible, the greatest number of possible avenues, independent of whether they would like them or not, would be a more direct way to increasing freedom, as opposed to trying to guess and... interpret or predict what that person desires or doesn't desire, you know, independent of whether that's good of them to desire that or not. Yeah. So what I would say to that is, we spoke just briefly about this idea of those two bubbles.
James Brusseau: Yeah. I said there's a freedom square, which is this idea of choosing to choose. And then what you want there is just that, if you go to a place where you've never been, you can see what people have options that are unimagined. So I would call that the higher level of freedom or the lower level of freedom. It's just, I want coke, but I want something else. But going back to... So I agree with you strongly that there's that division in freedom and desire. But then going back to your basic question, which is, is there some other way to think of freedom? Yes, certainly. There are many ways we can think of freedom. Just in physical terms or in terms of... Some people think of freedom just purely mentally, like what you can imagine you can't imagine. Those are all... When we can think of freedom, some people think of freedom, they want to think of it socially. So they say, well, what does freedom mean not for me, but for us? And then they say, well, the definition of freedom socially is, I can do whatever I want up until the point where what I'm doing interferes with you. And so that's a different way of thinking of freedom. It's just a grammar question. But I grounded in human desire because my experience of artificial intelligence, and I think the experience we're going to have of artificial intelligence is highly, highly personalized. That's what AI does. It used to be that companies designed products for 50 to 60-year-olds who had an income of $100,000. Now, companies are designing products for a changed person. Personalization is what AI does. And so if the world today is about personalization and freedom, at least the way we understand it, then the way the world is today is going to be about me. And once it's about me, I think it's going to be about what I'm doing. So I think that that way of looking at freedom, you know, personally, that's a narrow way. However, I think it's an appropriate way today. Because of artificial intelligence and personalization, that technology is one.
Other speaker (context): Just bouncing off of the last thing that you were saying. So you're basically saying like that there's a variety of different ways that we can think of freedom, but because you're working within the domain of AI, it seems as if AI is unavoidably, it's just this particular kind of freedom because its whole purpose is to give us what we desire, right? In other words, even if you personally as a human might have a preference for a type of freedom that has to do more with, let's say, self-realization or, I don't know, or developing your virtues or something like that. Maybe you have a different perception of freedom. But if in regards to this particular research, you're saying that, like, we have to look at this desire fulfillment type of freedom because that's the only because that's just how AI works. Is that kind of what you're saying?
James Brusseau: I think I'm trying to say that like that yourself at the moment, I don't think so that has to be often just because sounds what if the self is being fulfilled? That self has to be already given by the society and by those people who have to ask for help. So to me at least that sounds like often just whereas by contrast the idea of freedom that I am opposed to, to those people who are opposed to this, is that there is no self to be fulfilled. The idea of self really doesn't make sense because there's no constant me there. And instead I'm just here to make sure that it doesn't change it. If there's any sort of fulfillment in self, it's what did I create for myself? That's what it creates. So the way that I respond to that is by saying I'm of course that would be a different way in a few years but it doesn't put anything but it doesn't put anything except for me that it's the ability to be who you are to be yourself. So that would be a different way you're maybe also there's a difference. I'm just kind of because of the way what's happened it seems to me that you should hide your self. It's appropriate to today's reality. What Heidegger said is convenience and authenticity is more about developing
Speaker uncertain: I think this is Other speaker (context): Similar stuff at the same question but the first Speaker uncertain: Example you gave of choosing Other speaker (context): Your major Speaker uncertain: You said that's a real transformation Other speaker (context): Because it might create a world Speaker uncertain: Or a point of view for you and Other speaker (context): And contrasted to that to like if you just chose a bunch of different classes right in that way that feels like seeking novelty in every case and avoiding Speaker uncertain: That Other speaker (context): That self-definition step Speaker uncertain: But Other speaker (context): But in that first example it seems like there's something other than just novelty Speaker uncertain: Or just creation Other speaker (context): That makes a choice good or bad or makes it complete or not and so then it feels like there is a stage of like self-building Speaker uncertain: Up that's Other speaker (context): That's contrasted in some ways Speaker uncertain: To
James Brusseau: Right, the, like, self-exploration or change, right? So what self-fulfillment is like, it changes if you change your nature. If your nature is—self-fulfillment is going to happen, it's going to follow. And it's like a real relationship with animals. I mean, you might understand that if your nature changes with philosophy, there's going to be self-fulfillment, because that does not have to be a pure idea. So I think that that's going to change. So self-fulfillment changes with the choices, and then the next thing...
Other speaker (context): Wonderful talk you gave me a lot to think about and Other speaker (context): In Deleuze talks about like Other speaker (context): What is it's actually Other speaker (context): The possible, the virtual and also the real, right? And so it's just when it comes to recommendation systems, they consume your past and then, with some level of, like, randomization or some, like, they engineer randomness in some way and try to inject that into their algorithms. And they present to you these possibilities that, if you click on and if you actually like, they become real in a sense. But I think what's very, very uncomputable about life, or, like, true newness in a sense, is the fact that, like, the virtual is, like, so sudden, it's so chaotic, it's so, I don't know, it's a moment of rupture that cannot be predicted in any way possible. Like Alexander Galloway talks about, like, this concept in Uncomputable, right? So I think whatever the algorithmic world would just try to come up with, I think they get kicked somehow.
Other speaker (context): Because that moment that maybe someone interested in critical theory might be spending I don't know his or their weekend going through I don't know luxury vinyl I don't know plates or they could be doing whatever it's not something that could Other speaker (context): Be computed in any sense, right? So I think it's sort of like, I don't know, I wonder, like, what is newness? And then when we talk about that, what is, how we can—or, like, your metaphorical example of, like, one-way ticket also reminded me of, like, One-Way Street by Walter Benjamin. Like, in the early days of internet we would just click here and there, like we would just go, and then internet was not full of signs, right? And then there were not ads, there were not... But then recommendation systems, like, surgically, I don't know, made it so surgically precise that you can actually, like, see a pattern, right? And then, so it just—I think modern systems, like, they assault your senses, and I think as a result, like, scrolling, doomscrolling on the internet, we've just come across something that we don't like or we think that it just requires too much... Other speaker (context): Cognitive load and we would just pass it and I think our way of like developing ADHD or like attention disorder all of that is just a way of like reacting to this to that newness, that engineered newness that these recommendation systems try to come up with what do you think about that
Other speaker (context): Sorry, I probably there's another slide
James Brusseau: Just let me start with this. So Deleuze has a small book which I think is very palatable for his life. The book I'm calling Toward a Minor Literature, it's called, technically the title is Kafka, it's about Kafka, but the subtitle is the real title. The subtitle is Toward a Minor Literature. And what he tries to show there is how—I'm a subject of a book like this, how do I do this—how a new language does not come from over the horizon as some, like, a lightning bolt, something like that. Instead Deleuze says that what is new or what is confusing or what you're talking about comes from within the standard of the traditions that you already have. So what he was talking about, this language, is how it becomes a minor language, how the language can develop out of a language that is a subject of a book like this, and how this language developed out of the German language, slowly but surely took on a kind of substance of its own in his writings in terms of stories and so on, which in essence overthrew, which ultimately appeared to be new in the way that you're talking about. But the point I'm trying to make is this newness did not arise because of some radically different origin. The newness came because of some little change that was made within the standard way of different things. And that's also what I'm trying to do with this. Let's get the standard way of doing it, just make a little change and see if that can result in something that's new. And I think that changes a whole lot of things. So I think the idea of what you're seeing right now, I think the idea was that his project was always to discover things. And I think the heart of his project is to look into the languages, the tools we have now, say, how can we make little changes within them to create something which is different? Kind of like how this literature was really a different kind of literature, but it developed very carefully within—it wasn't just like you woke up one day and had visions of some new language. It developed within you. So I guess that would be my response to what you want to see.
[Unintelligible speech]
Other speaker (context): Yeah I'm sorry I threw in a lot of thoughts into that it was just all over the place
Other speaker (context): Sorry about that. You can just ignore me I think Hamed's question was pretty good because I had a very similar question because as an AI engineer as Hamed we've been always curious about does the AI change the definition of novelty or creativeness or it is just mimicking our Other speaker (context): Our assets or something so you know for example when we prompt when we give the prompt to Other speaker (context): The ChatGPT that please write a new article or new novel, Other speaker (context): Very short novel or something, Other speaker (context): Then can you say that that is a really something novel, Other speaker (context): Novel or something? Other speaker (context): Or when we ask ChatGPT or any other AI engine to draw some new images, Other speaker (context): Then is that just mimicking the, you know, Other speaker (context): Because when we train the AI model, it's just training. Meaning we give the answer and then the AI should be trained with that answer.
James Brusseau: So that is something very fine line question. And then I was just curious how you think about it. No, I think that is very true.
Other speaker (context): I think, I'm pretty sure that Deleuze's instincts would be to say that the machine is not creative. And those are my instincts. But if you look at his theory, his theory seems to indicate that the machines can be creative. Because of his definition of creativity, he is rearranging what we have slightly in a way that kind of messes up the whole system. And that's his definition. If that's true, then it seems like AI should in fact be able to do that. That's what AI does for you. It just takes the paintings that already exist and rearranges them a little bit to make a difference.
Now, I suppose, I will try to speak for the listener, I suppose that he would say that what humans have is a different purpose than us. And the way of maintaining it has to be further and further until it reaches an extreme. And I'm not sure where an extreme or a constant radical difference. And I'm not sure if computers are able to do that if they don't know what radical difference is. But in terms of the structure of creativity, I think, I personally want to say no, think computers can't be creative. But your question forces me to admit that by my definition,
Other speaker (context): It looks like...
Other speaker (context): I think we've lost our students.
Other speaker (context): Yeah, right.
Other speaker (context): Let's say thank you one more time. Other speaker (context): And please, talk to them a little bit more and then we'll let them go back to this wonderful room in the lobby.
Other speaker (context): Right, Awesome.
Other speaker (context): Them