AI Abstractions with Olga Goriunova

Olga Goriunova rejects digital abstractions as mirror images of ourselves and reflects on why we concern ourselves with representations that aren’t concerned about us.  

Olga and Kimberly discuss how cultural imagination is shaped by technology; digital subjects as unnatural constructs; the distance between individuals and their digital profiles; banal categorization and subjective truth; how statistics and ML changed the concept of the ideal; the limits of digital subjects; extreme individuation and aspiring to become our digital reflections; how current predictions create future realities; why the ideal digital subject isn’t concerned with you; and thinking critically about what we desire and why. 

Olga Goriunova is a cultural theorist working at the intersection of technology, philosophy, and aesthetics. A Professor of Media Arts at Royal Holloway, University of London, Olga is the author of the critically acclaimed book Ideal Subjects: The Abstract People of AI. 

Additional Resources: 

• Aksioma: Institute for Contemporary Art Book Lecture  
• Olga Goriunova Academic Profile  

A transcript of this episode is here. 

Creators and Guests

Host

Kimberly Nevala

Strategic advisor at SAS

Guest

Olga Goriunova

Professor of Media Arts at Royal Holloway, University of London

What is Pondering AI?

How is the use of artificial intelligence (AI) shaping our human experience?

Kimberly Nevala ponders the reality of AI with a diverse group of innovators, advocates and data scientists. Ethics and uncertainty. Automation and art. Work, politics and culture. In real life and online. Contemplate AI’s impact, for better and worse.

All presentations represent the opinions of the presenter and do not represent the position or the opinion of SAS.

KIMBERLY NEVALA: Welcome to Pondering AI. I'm your host, Kimberly Nevala.

In this episode, we're pondering data subjectification and representation with Olga Goriunova. Olga is a Professor of Media Arts at Royal Holloway, University of London, and author of the fantastic new book Ideal Subjects: The Abstract People of AI. Olga, welcome to the show.

OLGA GORIUNOVA: Thank you so much, Kimberly. Thank you for inviting me.

KIMBERLY NEVALA: Yes, I've been very excited for this conversation. Now, you are a cultural theorist who works at the intersection of multiple domains and disciplines. Can you tell us a little bit about what those disciplines are and if you had any inkling at all when you were starting out that this is where your life's work would end up?

OLGA GORIUNOVA: Yes. Good question. So I was originally trained in philology and the very attentive study of text and language, in all its manifestations, as it slowly unfolds through history. So I guess that remains, in some way, the aesthetic of the engagement. That I look for those kinds of aesthetic and literary imaginations but in culture that is now so much shaped by technology.

So to give you an anecdote about my early career efforts, when I was finishing university, I got a job in a computer magazine. And the editor - I don't know why or how it happened - the editor asked me on our first meeting, so do you know anything about computers? And I said, no, nothing. And he said, so what do you know about? And I said, oh, literature. And he said, well, write about computers and literature.

So then I started writing. And that was the end of the '90s, where the search engines were AltaVista, and you could actually find things, unlike now, 20 pages of results. So I started AltaVistaing and discovered - Google didn't exist at the time- that there's a whole range of computer literature and art with computers and media. And so this early internet gave me knowledge that, for me, remains the ideal web.

And through that, I became a specialist in digital media through this kind of literature and art, then digital media. Then, of course, I do close reading of technical systems in a similar way you would do of literary texts. I use philosophy, critical theory, but yes, with a lot of attention to aesthetics. Yes.

KIMBERLY NEVALA: Well, I have to say, the book was really exciting in a lot of ways because it brings together a lot of threads when we've been talking about this trend towards hyper personalization and where that might go wrong. And really trying to rationalize and think about what it means as things increasingly go digital, including representations of ourselves.

And the book - and correct me if I'm wrong here - in the book and generally, you really reject the idea of these digital selves as being a mirror or reflection of us as individual peoples in the world. Is that correct?

OLGA GORIUNOVA: Yes. Yes. So thank you. You've done a very attentive reading of the book. It's very impressive.

So in the first chapter, I spend a lot of time arguing that these abstractions of people, the data points that are somehow assembled and made sense of, go through multiple processes of abstraction. Multiple processes through which they are made sense of for specific functions, for advertisement, or for different kinds of targeting. And the way they are generated, our digital subjects, out of these data points have a little bit to do with us. But also it has a lot to do with the ways these domains work, data of other people, the functions and purposes of the exercise through which our data is modeled, and we are predicted to be certain kinds of subjects.

So to claim this kind of indexicality, or that this something that's generated of our data is related to us, to do this is a lot of work. A lot of work goes into making that claim. So it is not at all just a natural occurrence. And of course, then I call this process or this situation a distance; that there is a distance between people and their digital subjects. It is a distance that is really packed full of processes through which data is worked.

And that also is a good thing because it also allows you to disagree that this is your digital subject. Or that you are captured or somehow completely embraced, or violated, in this abstraction. And it gives the political possibility for us as a society to think that there is not a coincidence, and there isn't really this kind of direct tracing of our subjectivity, of ourselves, in the digital subjects.

KIMBERLY NEVALA: Yeah. And so in the book, you break down digital subjects into really two types. Or, that is what I took away from it, and you're going to correct my reading here as needed. One is the abstract subject, which is this version of ourselves - sometimes we might think about them as digital personas - that are created using our data and bits, and then also in collection, as you said. And then there’s this ideal subject.

First of all, did I get that-- I may have blurred the lines here a little bit. What are the defining attributes of the abstract subject, which I think you were just describing? And then we'll talk about the ideal subject and how that relates to or differs from the abstract.

OLGA GORIUNOVA: Yeah. So you're correct, completely. The first chapter looks at these processes of abstraction through which digital subjects are made.

So I start with very banal and little exercises and very simple free services like Google Analytics and look at what is a Google Analytics user and the way it's represented. And you see that these are data points that are somehow assembled into certain categories, but the categories are quite funny. It could be like avid investor, travel buff, and everything is on a certain kind of continuum of uncertainty. Because what's counted for a category is the number of sessions when you are recorded as belonging to that category or you expressed interest in that category. So the categories may shift. And even age can shift, or gender, depending on certain processes of classification.

So I look then at history of classification. And then, of course, all data science makes these claims that they are not any more bound to all principles of how science used to work. They don't classify. They don't use old instruments. But in fact, you have these kinds of abstracted attributes that are somehow made sense of when they are put into groups. And of course, it's not cost effective to target everyone as individual. So you would have to be put in specific groups and targeted as groups.

And in that sense, then, there is no real individual in these processes. The data points are really small, and they are below the level of the individual. I don't know, having lactose intolerance, for instance. And then the groups and the aggregations of people together that are above the level of the individual. And then you have things like profiles of possible people for which data, even if someone's data is missing, you can use the data of other people to infer that there could be a person like that.

So then it kind of leaves the terrain of this kind of individual self that we imagine and cling to. And it becomes this real procedure of abstraction and aggregation and disaggregation, and you're putting people into certain kinds of groups. And you see, well, do they work? Or maybe we need to find something else about this group. And multiple kinds of sorting is done.

So in that sense, I talk about this as abstraction because it is, in some way, rooted in the old abstractive processes through which science worked. Ordering the world through categories and creating these kinds of tables that systematize knowledge. But then, with AI and machine learning, as we know, tables are vectorized. You have everything in vector spaces and then these kinds of patterns and clustering of data points. So then something happened with these tables. They become non-normative. They go crazy. And they lose the kind of normative capacity. But then they transform into something else, and they acquire different kinds of threat.

So I look at abstraction because it's what sciences kind of do. And I look at how this works technically but also how, when philosophers started thinking about the possibility of thought, especially with Enlightenment in 17th century, the kind of famous beginning is that you look back at yourself to contemplate yourself, to confirm that you exist. To be a subject, you have to look back at yourself, reflect on your day, reflect on your behavior, reflect on who you are, think about what you want.

But what I set up in chapter 1 is that there is a distance in the abstraction through which digital subjects are created and it also very curiously resembles the distance that we are very used to when we are thinking about ourselves. And the second chapter says, well, but no one really seriously thinks about thinking anymore. We know now it's all about desire. So why do we care about these abstractions? If they can just operate in the abstract space, what do we care? So then I introduce something else and that comes with desire and with the ideal subjects.

KIMBERLY NEVALA: So I'm going to dive into all of those components. One of the things - or some of the elements that really struck me – again because I think they're very illustrative in terms of understanding why and where we need to be cautious. Where it's OK to buy in or lean into some of these areas, and when it becomes dangerous, really, in some ways.

In reading the book, one of the elements that really struck me was this thinking about the differential between me versus I - I in the collective sense. So what I experience or what I think my own sense of self is versus the version of the self that is being projected back at me by the systems and in all these digital ecosystems that we're working in.

You talked before about personas and profiles. And you give this example of the very common Spotify experience that I thought was just such a simple and grounded way of illustrating that point. And I think probably also tells us a little bit about that differential between abstract and the ideal. So can you tell us about that anecdote as well?

OLGA GORIUNOVA: Yes. Yes, I have a number of anecdotes that I used to anchor different ideas in the book.

So the Spotify example - and this really also maybe relates back or forward to the question of the limit of digital subjects and abstract subjects and ideal subjects - is that this term "subject" itself is really also a problem because we over fixate on this extreme individuality of ourselves. And we are asked all the time to polish ourselves as these precious stones, to become uniquely us. And that is the foundation of good mental health or something. That you have to be really careful, attentive to yourself. Which, in some ways, is a good practice.

But the question is how and what and whether, in this process, this extreme individualization obscures the fact that people exist only in relation to one another? We become in relation to others. We only learn in social situations. Language, knowledge. all exists only in relationships. And sociality and collectivity is extremely important and central to the process of becoming a self.

So in the Spotify example, in 2023, Spotify released these eight portraits that people could be matched with in terms of their listening habits. And some of them were like hypnotist, I remember. And so in that example, I was very interested in this kind of extreme individuation. The system tells you what you are like and who you are like. Which is presented as an objective knowledge because they have data on every song you listen to. So there is a certain level of facticity and objectivity about it. So this objective, factual knowledge is presented back to you as almost an existential truth or something for you to act on as a personal kind of singular, subjective truth. So there is already this tension between objective truth and subjective truth.

But then, also, when it's presented, it goes through another subjective process because someone came up with these banal - sorry, Spotify - very banal characters, pictures. [LAUGHS] So this objective knowledge is presented through a certain subjective filter. And then, also, when it's presented to you, even your individual listening habits, for their analysis, they rely on data of multiple other people. They analyze environment, time of day, holiday season, time of year, all of them. So nature comes into there. So many scales of experience are all pulled together to create this.

And it's just a very good example of this tension, and also the confusion that people have, because on one hand, it is supposed to talk to your inner self and help you develop your kind of unique interiority. And it's done through the authority of objective knowledge. But then none of this is what it appears. The objective is not objective. The subjective is not subjective. And then they keep shifting. Yes.

KIMBERLY NEVALA: And it's really interesting because I was thinking about this a little. There's a lot of examples in the book, other examples that highlight this idea. Where I might find it really interesting to see which artists I listen to, and which songs, and which categories those songs are in. So that's already a categorization that is subjective and not objective, per se. Although saying how many categories of music did you listen to would be an objective count.

But another thing in the book that's really interesting is that there - and I think this gets back to where we're going to start to run toward this idea of how we start to desire to be this ideal abstraction - is that if there's enough points of commonality in there. There's just enough points of verification where I'm going, I'm looking at it and going, I am in no way, shape, or form a hipster. I don't know if that's a category or not, actually. I'm just making it up. I just know I'm definitely not a hypnotist, I'm pretty sure.

But if there's just enough where I'm like, oh, yeah, but I did listen to those songs. Or I do like that kind of music. All of a sudden, I'm like, oh, wait, maybe this is valid. Maybe this is my persona and who I am. And that, in and of itself, is both tantalizing and dangerous in my view. Would you agree with that?

OLGA GORIUNOVA: Yes. That's a very good point. So there are these elements that anchor these abstractions. And I would also say that this desire for recognition, or to be affirmed as a subject, or to be given a certain kind of truth about yourself, or to be accepted, is always ambivalent.

So of course it would involve those moments at which you feel, oh, yeah, it's probably me. And then maybe fear, oh, is that me? Oh, what if that is me? I don't want to be like that. And then for specific kinds of people or forms of life, it could involve complete undoing, where what you're asked to aspire to is so unachievable and impossible that you can't possibly match this kind of ideal subject that is offered to you.

KIMBERLY NEVALA: Yeah. And again, something that I think is, when I read it, I was like, yeah, that's true. It seemed intuitive, but it's not something that I think I had seen so clearly defined, was the fact that when we started to use statistics - and this was well before the current advent and heavy-duty pushes to use machine learning, AI, GenAI today, towards hyper personalization per se. But you said statistics changes the idea of the ideal. When we started to use these techniques, the use of categories set up norms. And the patterns that were now being established using these techniques set up ideals. And that phrase has just stuck in my head and gone around in circles. Can you talk a little bit about that and why it's important for us to-- what that dynamic is and why it's important to understand?

OLGA GORIUNOVA: Yes. I'll give you a two-part answer to that.

With the statistics, I found this argument extremely interesting. It's inspired by a book by Davis. And it's an argument that starts with this idea that the ideal before statistics, even with all secularization, the ideal existed on a divine plane or a certain kind of transcendental plane.

So the examples that are given in the book, and that I take also from Davis, is that for a woman to be-- for a statue of Venus to be created, 12 women would model. And it would be decided that one woman would have ideal hands and another ideal head, but you couldn't have a single woman possessing all aspects of the ideal.

So the argument is that at that time, the ideal could be distributed maybe a little bit amongst people but it couldn't really be found on Earth. So that also meant that everyone was nonideal and inadequate. And I make this proposition that it was no point really desiring the ideal. The relationship to the ideal was probably not one of desire because it functions very differently.

And then with statistics, statistics began systematizing, measuring, and modeling human bodily capacity, emotional capacity, intellectual capacity. And the mean and average and the distribution of cognitive capacity in the population, these are all operations of measurement. There are abstractive operations. They're abstractions, but then they create a certain form of reality that then starts acting on us. And then we start finding that we have to act also in relation to it.

So the example I give is that when, say, IQ is measured - or all school tests are like that, all ability tests - they measure the distribution of your capacity in relation to that of population. And tell you whether they are top 3%, top 25%, bottom 25%. Which should be, you would say, irrelevant to your unfolding of your singular, meaningful human life. But in fact, it starts acting as this abstraction that, for some reason, we're now starting to want. You want to be in the 99th percentile. And even if you can't change your IQ, you still think, oh, maybe my children will be very intelligent.

So this ideal, when it becomes an operation of measurement and computation and abstraction, it also becomes something that is desirable or tantalizing. And it also starts functioning as something that you have to orient yourself towards to become a better person or to become a better functioning member of society.

And also, of course, it feeds into all societal planning models and, yeah, plans. And it becomes something also that is used for societal good. So it also reformulates the relations between the individual and the collective in that manner. And the individual is related to the collective but in this way in which the abstraction acts both on the collective level and on the individual level.

So that is the story of how abstraction becomes-- generally, even, how is it possible that we start even considering these abstractions as something that has an effect on our life? And not just the effect that just kind of goes on us, boom, like this. We are violated by, I don't know, some government calculates something and says, oh, your street needs to go, something like this. So you can understand abstraction like that; that there was some kind of assessment and certain models are used and they produce an abstraction but then it has an effect on individual life. But how is it that we, from bottom-up, you could say, start dealing with abstractions in this way? That they start having an effect on us?

It's what I describe as desiring the abstraction and desiring the ideal. So I write that there is a moment at which ideal becomes engaged in the processes of calculation. And I think that is something that just kept continuing since that time. So that is what I suggest: that we start desiring these ideals that are produced through formula. And desire is a force. This kind of openness to the world and orientation to the world is never only positive. Within it, you always have also rejection or ambivalence. And it can be repressive. So it's not necessarily, yeah, it's not a good thing in itself.

So that's one part of the answer to why and how is it even possible that we have this kind of intimate relationship? We consider these results of calculation as something that can have an effect on us and that we should aspire to somehow reject or, in any way, take a position towards.

The categories set up norms and patterns set up ideals. So that phrase comes from the consideration that when you had categories and statistics and norms, you also had different processes through which things were kind of checked and managed. And statistics as a discipline has also its own tools through which things are controlled.

But also politically, I think, we have developed strategies to deal with norms. When the whole history of feminism is, you could say, a certain response to the norm of what a woman should be and a push back and redefinition of a norm of a category of a woman. So there's a long history of dealing with those processes.

Whereas what happens now is that when you have these data patterns and you have this abstraction of relations, they produce forms of reality that are new in their genesis. And we don't have political habits and tools necessarily to deal with them so easily.

KIMBERLY NEVALA: This would then seem to foreshadow, in a lot of ways, a term we often hear within organizations and that of the "ideal customer."

OLGA GORIUNOVA: So the ideal as a term comes from two sources.

One is maths, where you have ideal mathematical objects which are projections. So you have a system. And to explain how it works, mathematicians would infer that there must be another element in the system that then explains how the system works. And this projected element in third element is called the ideal mathematical object. And then it starts acting on its own at the next levels of analysis. It kind of acquires a life of its own.
So from that point of view, you have all these data points that are sorted in certain kinds of patterns and modeled and then projected into certain kinds of predictions. And you have abstractions of relations that then start acting as if they're reality and then start coming back to reality as if they described and captured its principles. So in that sense, it is in this history of creating inferences and abstracting something that then comes back and is postulated as the source or the true mode of existence of this thing.

And the other part of it is, of course, the ideal, as we use in common language, is something that is desirable. That something that is good, that something that we need to want or we want. And that ideal also makes an appearance here because all of these exercises, they set up goals. They desire outcomes. They set up conditions under which this would be considered a successful exercise or a failed one. What is the useful outcome? They set up the initial conditions and, I don't know, benchmark the models on the basis of how well they perform. And the whole idea of ideal customer, ideal situation, I first encountered in marketing when I started studying these processes in marketing. So it comes from this: what are the needs and what are the situations within which this exercise is set up?

And in that sense, on one hand, you have this kind of mathematical ideal and a certain kind of projection that then acquires a life of its own, becomes reality. But on the other hand, you have a situation where this whole thing is set up to function to certain ends and in a way that has specific measures of success or failure and specific evaluations and specific functions that are not really concerned with you, you know? The production of your ideal subject does not have you at its core. It doesn't care about you profoundly. Yes.

KIMBERLY NEVALA: You're right. The narrative is that these systems, and especially today, the AIs and your digital assistant, is going to reflect your desires back to you and help you achieve those. But as you're talking there, it has nothing to do, really, with you as the individual. The ideal that's being reflected back to us is the idea of ideal for whoever has actually created that projection. And they have an objective.

So we are naturally inclined to want to strive towards those evaluations. I may want to be a hipster. Never going to be. Never was. And for no good reason. I'm operating perfectly well, I think, without that. But that's an evaluation that's offered. And someone might say, that's what I want to be, because I think that seems better in the broader scope of things. And not necessarily just appreciating who you are and being comfortable there.

So is it also true, then, that in this case, that who is offering that projection and offering that ideal and what their intent is becomes really critically important?

OLGA GORIUNOVA: Of course, you're right. And maybe not necessarily even who, because it's not about specific people. There are systems and there are industries and there are whole markets that function to capture and trade this kind of possible aspects of people.
Like if you like watching a film or a TV, or if you like just relaxing on your sofa in the evening, watching something. That, in itself, as a little piece of characteristic, is then extracted and packaged with all the other people who might have similar inclinations. And then it becomes something that can be traded and sold and bought. And becomes the foundation of evaluation of, I don't know, a value of a company which offers these kinds of services. I don't know, Netflix. I'm speculating. But the thing is that you look at systems within systems and interests and different kind of groups of interests that then massage you towards the specific ends.

So some of it is like, OK, you have an evening, and you would like to watch something. But it's not necessarily linked to these different kinds of outcomes. That you have financial instruments that relate to the future projection you that you will always want this or that.

This promise, for instance, of Netflix, is that it would give you personalized viewing suggestions, because of course, it has to be cost effective. So you'll probably get a personalized poster. But the films are not very different, as people discover. So you have to look at, yeah, cost benefit. A lot of it is about the cost and trying to sell you something or somehow make a profit off you or direct you towards certain aims. That's certainly a very large part of it.

KIMBERLY NEVALA: And is it true, then, that a lot of times, maybe one of the effects, possibly intended, possibly not - probably so though - that when these ideal subjects, or these ideal personas and profiles are being reflected back to us, that they tend to flatten us into somewhat of almost a singular type of identity?

And it's based on specific data points, and specific actions in time that, again, don't represent the whole of us. I think you said multiplicity is not reflected in these predicted representations. And this is where that distance comes between you as a contrary person who might have desires and likes and things that just don't seem to naturally go together, or may just not make sense to anybody else, get lost in this abstraction.

OLGA GORIUNOVA: Yes. Yes. And also, I downloaded all Twitter categories that users are categorized into. I think it's freely available on the web. It's like 4,000 categories. And you look through them and they're incredibly banal. They are that you go and buy cheese in your corner shop. Or a lot of them deal with the fact that we have a bank card and that you actually buy food, or you buy flowers for some kind of celebration, and you have the propensity to do this.

There are no categories that are about the poetry you might like or the struggles you might be having, the ways in which you make your life meaningful. All of the things that poetry or literature or philosophy taught us is not there at all. And it's not even attempted to be captured because it's not of any interest, really. Even books, as a kind of diminishing market, is not really very much reflected. No one's really interested in offering and pushing you towards certain kinds of books because, I guess, the money is not there.

So of course, first, as a subject of consumption, there's a lot about getting people to consume more or consume new things. And that's very well documented. And there's a certain, then, banality, the kind of banal aesthetics of it. Yeah. I'm not sure if everyone's pushed towards the same, because the market is kind of producing this endless differentiation of tiny differences of objects, but there's a general push towards consumption as a form of self-realization that's been going for a long while. And now it's intensified with this as well. A lot of these processes are used for that: to model you as a certain kind of consumer and push you towards even more things. And that's extremely banal.

KIMBERLY NEVALA: Another element, I think, that's really critical in the book - and I believe is maybe the focus of the third chapter and final chapter - is talking about this designation of reality.

And you mentioned before, where sometimes these things get projected back to us and then they actually do influence - we start to desire them - so they influence our behavior, and it becomes an interesting self-propagating project. And somewhere in there, you might lose track of which came first. What you actually wanted or intended and then what they wanted. And it just kind of goes round and round. At one point, I think you use an example of the Mobius strip, which I thought was great.

But you had also said in the book that inference is often harmless to an individual, maybe at that point, but it's core to reconditioning the conditions - or to reconfiguring, sorry - the conditions of future life so that future rendering becomes reality. Can you talk a little bit about why it was important for you to really lean in and bring that point to folks' attention as well? Because I think it is incredibly important, especially at this moment in time.

OLGA GORIUNOVA: Thank you for this question. Of course, you know, I teach at university, so you would always have students who say, oh, I have nothing to hide. Why should I worry that my data is collected? Maybe they will recommend me something I would like. And the contrary student.

And the example I use in the book is that of a Biobank, which is a British organization that collects data from, I think, 500 participants over a very long period of time. So the participants signed the contracts in which they are volunteers. And the contract specified that this data would not be used for commercial purposes and for science and research only. And the tracking involves quite specific and quite substantial biological tracking. They, once in a while, do all kinds of tests, and they write diaries, and wear wearable devices. And the different kinds of biological markers are tracked over time in relation to their life habits and behaviors and choices and genetic predispositions and things like that.

So it all sounds good in some way. Except that it's been revealed that multiple companies that deal with pensions or insurance have been buying their data. And so you can then imagine that if a person gives their data to this organization, they personally are not going to be affected. Every individual data point is kind of meaningless and not very exciting. And that's what, in this book, a lot of these ideals and a lot of these patterns are really not exciting. And they're not good ideals. They are quite, as you say, banal. Who cares what your heart rate is at this day? What is this?

But at the same time, of course, you can immediately see that if people are of a specific ethnic group, or have specific kind of diseases that they develop over time in relation to specific genetic markers, or they have an occupation that puts them in a certain risk, then that would be used potentially in the future to determine the insurance costs and the pension provision for people who would be deemed like those people analyzed. So that is what is so strange. That these future people might not even be born yet, but they are already included in those calculations about the world that would then be structured and create a society of a certain sort.

And this is what I really liked - I read it in Adrian MacKenzie's book - when he writes that when you do the kind of statistical probabilities of the future, what happens is that you don't control the future. The future is still unknown. But you assign a percentage probability to every event that could happen. And so you know it, even if it remains uncertain. And these forms of knowledge which transform the options and the possibilities that then would inform how politics or social relations unfold are really terrible. Because you don't want to inhabit society like this, where people who are at risk would then be pushed even further at risk, through systems that are completely obscure and not known, invisible. And these models are never discussed in public or have public approval or any kind of oversight.

So I think this kind of generation of reality is an over imposition of one specific reality. Because you have this amazing kind of multiplicitous unfolding of the world and you also have very different kinds of being, subjectivity. Not necessarily this kind of one consuming subject. You have all forms of being at the same time when you - I don't know, maybe it sounds naïve - but when you lose yourself in gardening. Or you have a caring relationship which can really transform time. Then you also have experienced this different form of subjectivity.

But via these systems, everything gets flattened, and only one reality and one kind of subject gets imposed. Then you're like, OK, so that's me, or is that me? And you're either happy with it or unhappy with it, but you don't have many other options. You don't have to be happy with it, but you're then just stuck in this negative relationship to it. And the optionality is drastically reduced.

KIMBERLY NEVALA: It just sounds like this preserves this wheel we're on. Also, it just narrows the aperture, it narrows the opportunity, or it definitely doesn't encourage us, to think outside the box and to think creatively because it's all based on data and a certain point of time. So there's a context. There's both an environmental and social, a political context, in all of these things are also captured. It seems to be anchoring us to history in a way.

And I think that's always true, that history always defines or influences the future and how we move forward. But in this way. we're always going to be building something now that's always anchored in that. So that idea of new possibilities or a new way to do things entirely - maybe to look at how do we approach something like health care - isn't necessarily on the table when we're just looking to optimize within the system that we already have.

OLGA GORIUNOVA: Yes. You're right. But it's also not just history, but specific kinds of history, because not everything is captured. So you might also have something that is captured that you would already say, this is not my reality. Or I don't accept it even when it's captured. But then that's what gets reproduced. And then history, and the term "historical" should mean that it's changeable, but it becomes the opposite, that it gets fed into this abstractions, and then abstracted even further, and then starts acting as if it's reality.

KIMBERLY NEVALA: Very interesting. So as you are moving forward, from this research, what is it that you're thinking about next these days? What are the big open questions that you're grappling with and those that you think we should all perhaps grapple with a bit more?

OLGA GORIUNOVA: Yes. Thank you. In this part in the book which is about desiring abstraction I also talk about the truth. How there's this subjective truth, that is developed through the practice of knowing yourself, and objective truth, which is the truth of sciences or any formal system of logic or knowledge, and how these get reconfigured. And generally, I think this relationship between desire and truth and desiring abstraction, desiring truth, is something that I would like to explore further.

And of course, now, everything is about large language models. And you have to deal with your students, I have to deal with my students, presenting project ideas that are generated by ChatGPT. This is really clear that this is not their ideas. Minority of students, but still. And it generates in me such a rejection when the students present the products of LLM that it's clear to me that it's about desire. Because only where desire is involved can you have such an aversion to something.

And I'm thinking through this figure as well, about why we expect and desire the truth in text or the truth in a textual product. How did it come about that we expect or want or seek the truth from text? And I think this entire thing about ChatGPT - and people are talking to it and discussing things with it or rejecting it - is about this kind of relationship to text. Where we need something from text, or we seek or we position truth in it. And it is not our truth, but also it can be our truth or is related to how we should be becoming as people. This interests me a great deal now.

KIMBERLY NEVALA: Yeah, absolutely. It's fascinating. And maybe we'll be able to get you, to entice you, back on in the future to talk about where that thinking has led you. Who knows? Maybe another book as well.

OLGA GORIUNOVA: Hopefully. Hopefully.

[LAUGHTER]

KIMBERLY NEVALA: So as we sign off here, any final words or thoughts you'd like to just leave with the audience?

OLGA GORIUNOVA: I end the book with this proposition that we need to out abstract these abstractions. We need to create different ideals and use different ideals and aspire to different ideals than to those presented and created through these mechanisms. And I was thinking that, of course, arts and literature and different forms of culture and creativity always busied themselves with creating ideals. And we have a huge variety of different kinds of ideals.

I would say that defending the humanities and thinking about these fields of practice and knowledge that are busy with creating ideals is not a way out, but a way to survive in such a world. And it's clear also that people are using art and literature to do this. So, in that way, that's my homage to my mother discipline.

KIMBERLY NEVALA: That's excellent. And it's a fabulous, fabulous note to end on. So thank you so much for your time and insights today. As I said, I really enjoyed the book. I've gone back to it several times and I think it does a great job of really articulating some of the influences and forces at work in our world today. So thanks again.

OLGA GORIUNOVA: Thank you so much, Kimberly. Thank you.

KIMBERLY NEVALA: Alright. So if you'd like to continue learning from thinkers, doers, and advocates such as Olga, you can find us wherever you listen to podcasts and also on YouTube.