Éthique et IA : Comment rester humain à l’ère du numérique - Mathieu Guillermin enseignant-chercheur

Hello and welcome to Data Driven 101. I am Marc Sansem, host of this podcast, which is interested in concrete and varied applications of artificial intelligence and data.

In Data Driven 101, I receive professionals every week to share their experiences and their vision without filter. Mathieu Guirmin is a researcher in Philosophy and Ethics of Science and Technologies, member of the Confluence Research Unit of the UCLi, Catholic University of Lyon. Hello Mathieu. Hello. So Mathieu, can I let you introduce yourself, tell us a little about your career and your current work?

Yes, with pleasure. Well, I have a little particularity, I did physics, I did a doctorate in physics, but the goal was to do philosophy afterwards. I thought I would do philosophy of physics at that time. And then, the more I went, the more I discovered, the more I became interested in the philosophy of science in general and the questions of ethical technology, philosophy of technologies and things like that, and in particular the questions of digital technologies.

So there you go. And so now I'm a research teacher at the Catholic University of Lyon. I mainly teach a little bit of philosophy of epistemology sciences. So it's a little bit of the "what is intelligence or what is rationality?" But not artificial, for once. How can we do all that? And then, a big part of philosophy and ethics of artificial intelligence. I teach, I also do research on the subject.

Okay, so today you are in a project called NHNAI. Can you introduce us a little bit about these works? Yes, completely. So the acronym, I'm not going to pronounce it in English, but it's New Humanism in the Time of Neurosciences and AI, so AI in the other sense. And so, in fact, it's a project that

Well, from the observation that there are still societal and ethical issues quite important with IA and with neuroscience, with cognitive sciences, with, let's say, the sciences and technologies of intelligence a little bit. And that we cannot necessarily be satisfied with a sort of management of these problems, of these risks and regulation that is descending.

from the start of the experts, the political decision-makers, the great entrepreneurs, very powerful

who have the power to make concrete decisions, who can impose their standards. But there is a need for that, of course, there is a need for reflection, a little like that, in top-down, but we also need to have an emulation that is a little more horizontal, a little more, as wide as possible. Because, well, already, I think there are subjects, but we will surely come back to it, there are political subjects.

in these issues, which are subjects, in democracy, the political subjects, the great political orientations, we discuss them between us anyway, we are supposed to, in any case. And then there is also the whole problem of how much even the experts among themselves would be able to say what to do.

I doubt it a little bit. Even if it would be possible, there is this issue of how you do to have an effective regulation. And we don't put a policeman behind every entrepreneur, behind every technology, behind every machine. It's absolutely impossible. So at some point, it's also if there is a requirement. And then it can be a requirement by soft law, by consumer requirements, even by saying, I want that, I would pay that, I would not pay that.

And so, for that to happen, there has to be awareness of the subjects and, I think, a reflection as broad as possible by people who feel concerned at least and who want to take a little time to do that. And so the project is trying to respond to this need. We asked ourselves the question, as university students, how can we use the fact that there is a better reflection on ethical issues in IA? And the answer was at two levels. It was first

would there be a reflection to have on what it means to be human, in a bit of a global way, which is not, so we will directly confront such and such very specific ethical questions, first of all, but try to, through the subjects, think about where we are going.

who we are as humans, who we want to be tomorrow, who we want to become, and that includes... When I say "as humans", it doesn't mean that we ignore nature, it means as humans, living among other living beings in our environment, and so on.

And so that's a subject to work on, like a bit the idea that if you're looking to orient yourself in the jungle of an ethical question, you have a bit of the magnetic north there that gives you a reference. So that's important to discuss. And suddenly, the second level, which takes up what I was saying in the introduction, is that it's really something, this question, we have to talk about it between us. It's the political question. And so, suddenly, the second dimension, it was, can we help, we, universities, to put...

our resources at the disposal of a reflection in society on these subjects. And so, concretely, we are... So, it's under the banner of the International Federation of Catholic Universities. There must now be...

So, there are some in Chile, the United States, Taiwan, Kenya, and then a bit of a big pool in Europe too. And the idea is that each partner puts his resources in a network of people who are not in the same group,

Yes.

So that's the idea of the contribution of expertise. But then there is also the idea of organizing ourselves, places of debate, places of exchange. And so that's what we did. I think we might be able to talk a little bit more about it, but that's what we've been doing in the different partner countries since we started. It's been two years now that it's very active, even a little more in that sense. Okay, some kind of conferences?

Through which? Yeah, so it was a small conference format, but rather limited because the goal was to really offer people to express themselves. And so we even went through a little phase, but there was no crazy success. That's the joys of the POC. But because we are a little POC, even if it's a big project, it's still the idea of thinking about what should be done to make it go well in a broader way.

We tried to gather representatives of different groups in the society, each university saw it in its light, according to its context, and to start working on what questions would need to be discussed more precisely. And so we did that, and then we said to ourselves, the people representing groups will mobilize other people who will come to discuss and all.

It worked more or less, but it started a dynamic. And then, after that, we had a whole phase in 2023, mainly in-person events. The methodology was a lot of questions. It can be one question shared by the entire assembly or different themes that are worked on by 10 people. In any case, people were in small groups and tried to talk to each other.

And a little bit of originality, if you want, that's not very original. Besides, we are thinking about how we could have a little resources in terms of animation methodology. But on the other hand, a strong particularity was that the idea was to say, I will illustrate it like that with my hands, but can we not pass from agreement, not agreement to debates that would rather be on the

"Yes, but..." I often, in ethics classes with my students, when we have ethics debates, they say "No, I don't agree." "What was said before, you don't agree with?"

It's really no, no, no. Or it's yes, but... And very often, in fact, the person says yes, yes, it's not wrong what was said, but there's something else to say. And that's super important. And so we tried to really encourage people to go into this kind of logic of saying, let's not try to give an answer to the question integral, alone and unique. We'll see afterwards. Let's start by seeing if we can't agree on...

All that is not too stupid to say on one of the questions that we look at. So one of the aspects of being human, so much to link. And really with this idea of saying, I take the time to think and I say yes, but... And then maybe yes and sometimes too, because in fact it's yes and then I'll even go in the same direction as you. How did you choose these questions ?

So, the questions, we were very liberal. There were still three themes of the game, we will say, because in 2022, when we launched the thing, we saw that it was interesting. It was the field of education, the field of democratic life and democracy in general, and the field of health.

So, we had proposed that and possibly more transversal questions, a bit generic, in IA. And then, the questions were worked on by people locally, trying to interface with the possible participants. Sometimes, it was more precise questions on, for example, the brain-machine interface or human-machine or things like that.

because the local person had a community interested in that around them. Sometimes it was more generic things.

But in fact, it didn't bother us because what we tried to remove from these discussions in all the countries, it was really after a while, the level, what does it say about what it means to live, to be human and to live with others. And so finally, whatever the questions, the important elements that came up, they talked about that in the background. So we tried to follow that, but really with this methodology of saying, let's enrich ourselves.

-

I don't quite have this reading as a philosopher. I have the impression that you will not kill yourself, it is a priori always true. Simply, there are other things that are also true or sufficiently obvious. Besides, the style, you have the right to defend yourself, the style, you will protect the innocent who is threatened. And so sometimes it comes into conflict. And it's not opinions that come into conflict.

I mean, on fundamental values, there is no problem. However, what comes into conflict is possibly the priority that you will give to one axis rather than another. And so, that's why we can really try to make agreements on these ethical and political issues, at least by trying to say everything that is to be said that is not too stupid. And we see...

-

The way to cut it will depend on who prefers what. But it doesn't prevent us from agreeing on the existence of the problem. And that, me, I, by expanding a little bit, I think we are a little bit tired in society and our political life, it is sick to not know how to live in, also in this level where we agree on the minimum that holds the road. A little bit completely against what we are seeing in the United States there, with the side, everyone says what they want in their corner. And

And I don't necessarily want to forbid people to talk, but maybe it would be interesting to try to make the effort each to admit everything that is not stupid from the moment when it does not engage us to say that this is the solution. It's just that I recognize it, it's not stupid and I recognize the tensions.

And that would be really... Well, we really tried to do that. And somehow, the project, it tests this kind of approach a little bit. Can it take... And maybe even that the IA could... I mean, there are IA technologies that could help us do that better. We've been thinking about it too. So, these famous nodes of complexity, there is a list of subjects identified. Can we go into the details of these...

these opinions that are in tension, whether they are yours or not. Can you give us a little bit of an overview, a visualization? And maybe the one I would like to take first is that of the workplace and automation, which is a subject that does not only concern AI, which, let's say,

already have anecdotes dating from the time of the Romans on automation and the destruction of work that comes with it. Can you give us a little bit of an overview of this debate? Yes, of course. And so just a little clarification. In fact, I very honestly, but

Sometimes I have problems with my loved ones. I am. There are all opinions, I hear them. Everything we transcribed, I hear it and I think I don't think I filtered it. It's not me who did all the syntheses, by the way, but I don't think I filtered it under pretext that I didn't hear. And very honestly, everything we put, I hear it. And then, on the question of the automation of work,

What has been discussed a lot is the legitimate side, dimension, but of the improvement of efficiency, whether it is its own personal efficiency to do more things by itself, but also to be more productive and therefore by extension the legitimacy of wanting to be more productive and more profitable for a company or for an organization, for example. This was one of the elements that was recognized as being legitimate.

After that, it will come into tension with other things that have been expressed, such as the risk of finding oneself useless, which is not so obvious when you look at the reality of things, but which makes people talk a lot, given the great media announcements and probably to the goal of capitalization too.

But that's also expressed by concerns, saying: "I want to become more efficient, but if it makes me useless..." And then afterwards, it's an organization without

without me or without living inside, which is more efficient. It's a bit of a question. So there was this dimension, of course. There was the dimension of the risk of losing skills that we wouldn't want to lose. So for two, two, two, two, two titles, it would be because we might need them tomorrow and that in fact, we automated. And then finally, it was a brick

that we would have needed to keep. It was painful, maybe even, but we needed the know-how to do other things we didn't know. And so, in medicine or in things like that, we can see the thing coming. Because there are magnificent diagnostic algorithms on everything that is medical imaging, things like that. It works great.

So, it means that there is... You have to be careful about the arguments of saying that we are going to lighten the burden for the collective, for example, of the training of doctors or the number of doctors in the hospital or things like that.

But there shouldn't be any new subjects, new diseases, new problems where we don't have the examples and where we'll have to recreate examples. Well, unless you believe in things like J or who will...

We won't need to be told things before. And that, apart from cases a bit like Go or something, where you can calculate why it works or not, it's easy. So you can let a program look for better things without you. But in cases where the problem is not very well defined, in general, it is the humans who manage to go and see if it works or not. So here are some problems with deskilling in that sense, of loss of skills.

And eventually also problems of being forced to stop doing tasks that we liked to do. And there, there can be a real tension with almost the side, you don't do it enough efficiently, but you still want to do it. What do we do? So what we liked to do, sometimes just...

-

But it could also be a way of breathing during the day, of working, of not being... Honestly, I'm telling you a little bit of everything I've heard, everything I've read in the synths. Sometimes we've been a little more selective. But what you're saying reminds me, it was a little bit about these stories. I don't know in what context, but I heard it. Yes, with this side...

So if we only find ourselves with very, very high added value work, in these cases, we may have to work less. And suddenly, the gain in productivity is not so obvious as that, because you can't... Well, maybe with the evolution taking place in a thousand years, we will be able to be at x thousand percent of cognitive work. But there are times when it smokes, you have to stop too. So after that, that's it.

Just, that's where we see the whole political question of the subject and that it's neither you nor me, nor an entrepreneur alone in his corner. It may be eventually giants like the very big tech companies or countries, but

I have the impression that it's still a political issue, that it can be played out. But there is the whole question of the organization of work, of why we work for how long, where is the money, who benefits from it, things like that. And that, well, that's a little bit of me who adds it, but it has still been mentioned a little bit. You also have the justice side,

social and economic, on the fact that the data comes from where. We learned to do things, it's good here I can say it without explaining what it is, that learning, it changes me. But we learned to do certain tasks, to make the machine do certain tasks. But where does it come from? How was the data obtained? I tell myself, the people who wrote in Wikipedia at the time, they signed licenses that legally

to exploit it to make a great language model or other stuff like that. If they had been told that there were great language models, would they have contributed like that? There was an idea of common good and potentially there is a... So why not too? But it raises questions. I don't have the answers, but it raises these questions of how we want to share the gains that are produced with these things and with a real legitimacy to the

to all the people who develop and participate in the development and construction of computers, to say that it costs a lot and we have to live too, and it's legitimate. But there is this kind of issue. There is another issue that comes up a lot, the preservation of democracy, in particular the one of freedom of expression and regulation.

What is it about? I think that in particular, culturally, we have a fairly different approach from the Americans. Today, we see Trump, then Zuckerberg, coming back to something much more, ultimately, faithful to the American doctrine of everything is possible, everything is allowed. And we are almost surprised. What is it?

What is the state of the opinion on this? Well, also our little precision, our goal, it has never been to claim to do instantaneously what people think. We are really in a logic of construction in the discussion. I still specify because it represents only that. But it has risen and it's not surprising. All I'm telling you is things, if we are sensitized to the subject, it comes quite quickly. But it's the side...

There is a real... It's not a dilemma, but a real difficulty. You have to regulate, but it's very complicated to regulate.

Who keeps what, with what criteria and with what legitimacy? And we can see that there is a return of the baton in the United States. And then who will keep the guardians? In the project, what we tried to do was capture the complexities that are built by people when we put what has been said end to end.

or when they discussed between themselves and after the idea was to say can't we make an contribution of our expertise which helps to

explore these complexities. And because there I was going to say something that touches me a little more as a philosopher, but there is a moment, there is an illusion of "I am rational, I say legitimate and true things because I have objective criteria, I did my experiment, I did my calculation by a plus b, I have an absolute proof, there is no, in quotation marks, there is no problem of freedom because it is an absolute proof.

And so you don't have to choose, it's like that. And there is a bit of the idea that scientific stuff would be that. While in fact, it's not that. The reality of everything that we admit and that we recognize, we, you or me, we submit to certain authorities by making them trust. That's what's actually happening. And we trust them because they are supposed to do their job more or less well. But in fact, there is always a scientist. Well, I was in physics. You always have a part of judgment.

And when I say judgment, I'm talking about something, after a while, it's no longer with criteria. It's yes, because anyway, okay, I put this criterion there, but why this criterion rather than another? Well, because, and then after a while, it's because in my guts, I feel that this is what needs to be done. And that's a job, it's a failure, it's... We fall all the time, we have to stay vigilant. But in fact, that's what human intelligence is fundamentally. And so, for regulation, for moderation in particular,

It's not a flaw that people have to judge. It's inevitable that people will judge. There will never be criteria that will solve the problems. And on the other hand, the question is, OK, the judgment is feasible, but it's not because it's feasible that it's necessarily...

Hmm.

But yes, but it's normal and refuse to judge. It's not to be freer, in fact, it's to do. Everyone will say what they want in their corner and we will see. There will be no more living together. And in the end, that's less freedom. Because if you run the ball, we come back to the strongest on the weakest or the greatest number, the most convincing is right. But that's not freedom. It's it's it's it's at the limit, a lot of freedom for the strongest and the most convincing.

But otherwise, collective freedom is to judge together, to make this effort to judge together. And so behind these questions of regulation, sorry, that's why I said I would allow myself a little contribution of philosopher on the subject. But there is this political problem of not saying whether it is necessary to regulate or not, but to think together how and to accept that it is not perfect, that it is done by people who live the thing and who do their best and that afterwards we are almost on the quality of people and the trust we give them.

more than on such and such criteria, which would be a bit of an illusion. Another subject which is much more niche, let's say, a little less...

On the top of the pile of ethical subjects, but very interesting at the level of education, you have mentioned it as how to make quality education accessible while preserving cognitive development. Can you give us a little bit of the subject? Yes, I can give you a few elements. It was the idea of ... what has been said, everyone has not said everything, but the idea is that we build together.

There are plenty of technologies that can be super useful for school and especially in terms of inclusivity. For the things that allow you to do sign language, for example, not at the fly, but soon it will surely be the case, or subtitling or things like that on videos, it can make it accessible to lots of people. All the knowledge now, even if it's not in your language, you can still have access to get a little bit of a mess. You have great tools for learning mathematics, stuff.

a bien utilisé, justement. Tu as tout le côté personnalisation qui permet d'espérer...

Once again, I immediately want to go from the other side, but I'm going to finish on this side. There is a real recognition that yes, children have needs to personalize and that potentially, if you can know who needs what, at what time or if there is a little more flexibility than in a wine class where the teacher says about everything, the same thing to everyone. It's interesting. Nevertheless, behind there is also the side. Yes, but be careful because there are several, several dimensions, but there is a dimension

you can't have the algorithms do it instead.

because it's not at all the same. If assistance becomes a replacement for reflection, it doesn't work. I'm always fed up with what our students say. I want to confront them all the time at the university. As soon as they have the opportunity, they do a job with GPT or something else and they are quite expert in using that. And I'm not saying at all that you shouldn't use it in itself, that's not the problem. But what's bothering me and worrying me is

this side, total absence of reflection on why you are there. With a kid in the CP, he doesn't know and he fails. But at the university, even in master's,

I don't know, you're here to learn something. If you do it by machine, we don't care. At the end of the day, she's going to do it well, she's going to do it badly. You'll have a good grade, you'll have a bad grade, but you won't have experienced anything. And that, I'm not throwing stones at the students because it's a whole social-historical thickness that makes that when it started to be accessible to everyone, people at school, they were: "Ah, that's great, I can go to school."

And now it's become a bit, it's a bit of a mess. We don't know why we're going there anymore. There is no longer this culture that is a bit lost too, because I don't know, at school I lived it as very hostile too. So, so there you go. So there is all this problem of, it joins a bit the side of skilling. It's not the idea, it's not to delegate, what? Make more accessible. - In what measure is it different from what my generation has known with research engines, for example?

-

So there would be that, but otherwise, in the idea, we are in the extension. Yes, it can happen. It didn't happen with the copy-paste. Yeah, well, no, if you do a copy-paste without rereading, it will be seen. And that's it. And so that's it. There is this side. It's not made to make life easy.

In fact, wanting it to be easy, it's a bit of a dirty problem too. There is a moment, if you want to learn, and we tolerate it better in sports. It's normal to do abs and do exercises where we're going to get drunk. If we want to be high level athletes or even play a sport well, if we want to do an intellectual thing well, it's the same. There is a moment, you have to borrow the paths, you have to do things. It's the little task that we don't do anymore because, oh, man, we needed it. So...

So there is this problem that joins the side a little bit. So there, it's not deskilling, but it's the absence of getting into skills or learning. But because you don't see anything, you haven't asked yourself the question of "I'm here to live what?" And suddenly you don't see things that could have been interesting. After me, I have a personal title, but I express myself in this debate too. After all, my students, I tell them, at the limit,

I know they don't listen to me, but not all of them. Prove to me why the exercise I'm asking you is useless.

I will listen to you seriously. And if you're right, thank you, because you will allow me to think about a better exercise and I will give you a good grade, because it means that you have understood everything about the objectives and it's cool. You have lived something interesting. My goal is not to make you do my exercise for fun. My goal is that you live something that allows you to develop yourself internally. And just the text. But that's what we're observing.

We have a correlation, even stronger than before, and really radical now, between the possibility of having a convincing end product from a pedagogical point of view,

of a pedagogical exercise, which is completely uncorrelated to life and the experience lived and the inner movement that there is in learning. And that's the same for researchers who write papers with GPT. It's good, but isn't it filling goals that are not the goals of a community of researchers? It's kind of the same idea.

Yes, on the education background, we could not, I may be a bit optimistic about it, but finally wait for a rise in maturity thanks to this very strong exposure, this absolute ease of writing the answer in one click, to actually rise in maturity and say: "ok, but why am I doing it at the base?"

Yes, yes, completely. I often say that, I'm a little more on the personal register, well, as a researcher, but I often say that, I think and I hope that it's an opportunity to talk about it. And in some way, the problem, it was a theoretical problem before.

There was a time when if you wanted to have a good grade, you had to work a minimum. And now, in fact, you can make products that deserve a good grade without working. And so it allows to, somewhere, to put the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work back on the question of why I would work on the question of why I would work on the question of why I would work on the question

We don't even look to know if it's true or if it's false, we stay in the benefit of doubt permanently and the effect of fake news

-

Maybe it's those who already have a lot of social capital, parents who can frame and so on, and that the others will not have the opportunity to think. And then, what you point out is super interesting and it's a challenge. It's extremely dangerous. We need a factual common world.

So, it's sure that we... Me, it's my part in the field of science, I work a lot on things like that, but it's sure that, in fact, we discover that we are never completely sure of facts, that we can never have criteria that will suddenly tell you, that's the truth. But that has always been true and we have lived with it until now. Sometimes, we were not told completely true things, but we must have a factual world that we do not question permanently. Because if we don't have that,

So the consequences can be monstrous. Just to explain that, to be able to

to project oneself, even to have a political life, well that's Arendt, Ricoeur, those are really things that they studied after the Second World War and the whole problem of extermination of Jews and all, to be able to, and it's 1984 if you want, those themes, to be able to project oneself in, I would like the world to be like that, you have to start from somewhere solid,

You have to have a shared recognition of the world, it is like that. And so, I would like it to be different, because like that, it poses such and such a problem, and if it was different, it would be better. If you denounce this relationship, ah, the world is like that, a little fallible, we discover it more and more, but still, what. And this will to recognize a common world,

you kill this ability to have a political life behind. And there, for once, we almost all become people who are being walked around because... But the whole message of 1984, that's it, it's to prevent common sense, to prevent existence... I think it's the crime of absolute thought, it's common sense, it's to say, but still, well, no, freedom and submission, it's not the same thing, or I don't know which couple they take, but... And the party, it wants to keep both together so that people lose the habit of judging each other.

lose the habit of making the effort to recognize something in common. And that kills any capacity of political opposition or political movement. And then you become almost an object who suffers what happens to him concretely without being able to talk about it, without being able to bounce back on it. I close the parenthesis a little bit, but the consequences of what is happening in the United States are worrying. The rest is not going either, but we come back to what we discussed a little bit before. It's...

saying "everyone will make their own opinion" will lead to everyone saying what they want in their own corner. We will very well accommodate that there are alternative versions of things and that, in the end, leads to no longer being able to have a world on which we rely together to talk to each other at least about where we want to go after.

And it's an illusion of freedom. It's an illusion of freedom of expression. Because yes, if you have the right to say the words you want, but to express yourself, is it not also to engage in something of yourself, to engage, to try to tell someone else? I think that's what's good. And I think I'm offering you all these things. You lose the freedom to do that, in fact, when you have the freedom to say everything and anything, no matter how. But I don't even talk about regulation.

with, how to say, punishments and people who control. But just to say I have a responsibility, not to say just what's going through my head because it suits me, but to make the effort to validate what I say, why, and to have this will to validate together. Yes, so the United States often had the responsibility at the level of

des sociétés, et notamment des journaux, la télé, on ne peut pas dire merde à la télé, il y a un bip, ce n'est pas la loi, c'est la chaîne qui refuse. C'est souvent finalement, la censure est venue des sociétés, aujourd'hui, du coup, il y a un peu, finalement, un refus de jeu de leur part, c'est une sorte de...

Yeah, game refusals or... But there is also... And somewhere, I think it was dangerous... Well, it's dangerous that there are gatekeepers too concentrated who make the content. And from that point of view, I don't know if Zuckerberg is really philosophically engaged in it and all, but there is a... There is somewhere an interest in having a liberalization

information space. The problem is how you organize it. And because the gatekeepers, they did so, if you're lucky, you have a team of journalists that you think works really well, you will be able to trust them.

Otherwise, you know it's biased. Me, even journalists today, I try to read a little left and right everywhere to have ... Because here they are, they are not necessarily unhealthy, but they have their filter. And so it's very good that it can liberalize itself and that somewhere there is not too much concentration in who moderates. But on the other hand, it raises the question of how we moderate and how we recommend. And there we are going towards, we will put notes and we will push the contents

we will prohibit poorly written content, but who votes, by virtue of what? And especially that in fact, I don't want to make an intentional trial, but I think that for once, the absence of regulation on social networks, that's what Gerald Brunner, Apocalypse Cognitive, is about. He's a sociologist who wrote this about saying: "It's not because we're nice or mean, but because of our biological evolution."

We evolved to pay attention to certain things and less to others and for survival reasons, reproduction, food, things like that. And so what catches our attention is often the content a little outrageous, provocative, the clash, sexuality, things like that.

And it's not necessarily what is ultimately the most fulfilling, but that's what catches the attention. And so, if you deregulate and you say, "Everyone will say what they want", it will go in the direction of favoring these contents. And while potentially you would have wanted to calm the game a little, but to calm the game a little, it's in quotes a little bit, "Oh, well, it's boring. It's boring, there's no clash, it's consensual, people talk to each other, they listen to each other talk, my God, for ten minutes.

But in fact, our brain, it may not have the patience and we are, we may be shivering when we have things a little bit in clash mode or in mode, it's not going to get too far. The problem is that it's this bunch of strawberry tagada. At one point, we might want to go eat at Bocuse or someone else's too. It's an interest. And the brain, it is ultra happy. And then the more you give him, the more he wants it. And there is still something to know about ourselves that raises questions.

But at the same time, it doesn't mean that it's a bad thing that we have spaces where everyone can put what they want. But there is, after that, we come back to the problem, I think we talked about it before, on the problem of the economic model and recommendations that are, you don't pay.

you don't pay for your informational content today and not bad tools either. And in fact, there is a moment, it must be remunerated and how it is remunerated by generating interaction with the digital system of data. So there, I'm going to learn nothing to anyone, but to personal data, possibly, but it's actually a false subject. But profiling data, even anonymized for advertising or content recommendations,

either advertising because we pay advertising spaces or content because it reinforces the commitment of people in the system and after that leaves data that is potentially bearer of examples of tasks and there for the ia it is the big business to collect high-value data added that will allow to make better models tomorrow and that's the model and therefore the information if

If there are no other counterweights and other sources of income, it will necessarily be organized towards what generates the most interaction with the system. At the limit, no matter the content, in quotes. Still, you could say to yourself, if I want to make super sharp code, I may be interested in having spaces where people do things with a little higher added cognitive value. And so, if they spend their time watching cats running after themselves, it may not be the most interesting thing.

But as a citizen, as a consumer, we need to think about whether it's such a good thing not to pay. Also because we consume, the sweets are free and I go. If the sweets were paid, I would say, what do I buy? How much does it cost me? I may put aside to go to another restaurant. If I follow the metaphor. Can we not expect that somehow, in the same way that...

Transformed sugars have arrived, nobody is suspicious of it, there has been a huge obesity in countries that have pushed it more precisely. The metaphor you make is very interesting because it has already happened. Then today, we have taken a step back, the same with cigarettes, I mean, the same subject, we have taken a step back, we have taken measures to inform people about the side effects of cigarettes, of excess sugar.

And today, we know that these things can be a pleasure and that we have to resist them. And can we project ourselves into a world in 20 or 30 years, in the middle of generations sacrificed, like every time, without a doubt, where everyone knows that taking their phone and starting to scroll is killing themselves. I wish, I hope so.

And somehow, the project as I designed it, it's to help with that, clearly, because it has to be done by everyone. It's not a big regulation. It can sometimes, but it comes after a year because there are problems. People see all their kids. There are too many obese kids or things like that. So there's a scandal and it moves.

So yes, of course. And me, by the way, from that point of view, I really fight to break this idea. But I think that in the middle of entrepreneurship, every time I talk about it, people are super agree. It's this phantasmagoria around the IA that advances on its own.

And a wave like that, which does not depend on humans, it's super unhealthy. It's not that. When you know what it means... When I hear "we are losing the war of artificial intelligence", yeah, we may lose the war of the development of the biggest models, which we don't even know what we will do, because...

So maybe they will be super disruptive later, but I don't know. But what I see on a daily basis, is the question of the IA, it's the question of who puts what type of code in his daily life to gain productivity, to do more interesting things, to live better and so on. And that's human work to do. When you're an entrepreneur, you work in it. There are lots of people who work, who do great jobs for that. And so, in a way, reality is still, and it's not healthy to hide it, because anyway, it's human choice.

What we will be with our information systems tomorrow, including AI, is what we will have chosen to become collectively. After, the question is who chooses, who suffers, who is aware that there are choices to make, and so on. But if we improve the level of awareness that there are these subjects, and the will and the ability of people to participate in this reflection, I am...

I become an optimist again, in quotes. It's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's it's

So, if there are avalanche effects, they can be ultra-powerful. Simply, we have to trigger them and trigger them in the right way. And so, no, no, I think we can. Besides, there is still a game with certain people who are still in the middle of regulation, who are responsible. And there, we start talking about problems of epidemics, of screen addiction or stuff like that. We can see that it poses problems, it becomes obvious. And so, that creates subjects.

And we talk about it and so it creates the conditions for... But on the other hand, really say, these are political subjects. And we can't say, get rid of the people of tech, it's your job. And we can't let certain people of tech who want to capitalize on it. Don't worry, dad acts or mom acts, AGI arrives tomorrow and it will fix everything. All that is irresponsible. We really have to say...

We all have a responsibility. We don't have to be involved in all the subjects, but they are political subjects. If it doesn't go well, we can't... If we don't have any political life at all and we constantly whine about what's going on around us in the world, we're not coherent. So, not everyone is going to get involved in all the subjects, but if everyone got involved at least on one subject that they care about, we would start to have this ability to create a somewhat rising emulation, the one we're trying to contribute to.

So do you have an invitee that you could suggest for Data Driven 101? I would encourage, I was talking about Gerald Brunner who does these topics on the question of attention time and recommendation algorithms. It could be someone quite nice to hear. I don't know if he would be willing to come back. But in addition, there is a need for this link between people who are like us. We are often told we are out of the ground. Me, my friends, when they saw me do my physics, philosophy studies, they were like

"What are you doing?" Well, after that, I had more and more university friends, so we're still friends, but it's less difficult for the ego. But they're a little right in the sense that sometimes, the papers you publish are read by ten people, and then your university is happy because there's a paper that's done, and then, in fact, the link doesn't make sense with the rest.

And it's not easy to do that and to have places like the one you organize and that you animate where we can also exchange, not only between business entrepreneurs and us, we will talk in our seminars, but that there is also a link with people who have a slightly more theoretical approach, certainly, but who...

Thank you very much Mathieu. Thank you.