Shaping Ethical AI (English version)

There was a time when people said: “the Americans invent, the Chinese copy, the Europeans regulate”. Today, the United States dominates AI, China makes the most advanced products… and Europe watches the train go by while debating whether ChatGPT should say “hello” in gender-inclusive language.

Europe is quietly losing the global race for artificial intelligence and innovation. The United States invents, funds and dominates: more than 100 billion dollars invested in AI in 2024, against less than 10 billion for China and a few crumbs for Europe. China, for its part, no longer copies: it now controls global production of robots, batteries and semiconductors.

Meanwhile, the European Union keeps perfecting its regulatory arsenal, with no industrial strategy and no shared vision. France likes to dream of itself as Asterix’s Gaulish village: Mistral AI shines, Paris attracts, but 56% of French capital flows out to the United States, and the talent follows.

The result: a continent regulating what it no longer produces, dependent on American clouds and Asian supply chains. If we are not careful, the risk is to become a digital colony for good, a mere consumer of technologies designed elsewhere.

Europe can still react, by linking regulation with industrial strategy, by investing massively, by unifying its markets. But the innovation train will not wait for those who stay on the platform arguing about the colour of the carriages.

Find the full transcript of this episode, along with all the sources and plenty more, at: The silent decline

344 times a day. That is the average number of times an adult checks their phone, according to the BBC. Every four minutes. Even at night, even for no reason, even without a notification.

In tropical forests, there is an ant that climbs onto a leaf, clamps itself there and dies. A fungus has taken control of its nervous system to spread its spores more effectively. Biologists call it a parasitic infection. Poets would call it a dispossession of the living.

That fungus took millions of years to perfect its strategy. Our smartphones needed only a decade. The biological parasite forces its host. The digital parasite has convinced its host that it can no longer live without it.

This parallel between biological parasitism and digital colonisation is not just a metaphor. It is a modus operandi. The same mechanisms, the same precision, the same result: a host working for its parasite while believing it acts freely.

So perhaps the real question is not: are we addicted to our screens? But at what point did we stop being the masters and become the host?

Find the full transcript of this episode, along with all the sources and plenty more, at: Willingly embraced parasitism

This is the third and final instalment of my series on the privacy of our data. Having explored our own surrenders and the illusion of voluntary transparency, it is time to ask the most unsettling question of all:

What are we leaving our children? Not a material inheritance, but an inheritance of outlook.

For they are born into a world where intimacy fades before it has even existed, where surveillance dresses up as play, where freedom is confused with being permanently connected. What was a loss for us is, for them, a given. Where we see an invasion of privacy, they simply see life.

This podcast examines that silent shift: how do you pass on inner freedom to a generation that has never known secrecy? How do you teach depth to those who have been taught to expose themselves? And above all, what will be left of freedom if we forget to teach it to them?

Find the full transcript of this episode, along with all the sources and plenty more, at: What are we passing down to our children?

We believe we are sailing free, but we are moving through a strange bestiary of myths revisited. Like Narcissus, we lean over the digital mirror, captivated by a reflection that ends up swallowing us. Like Sisyphus, we carry the burden of a memory that never forgets: every piece of data is added to the rock that crushes us and never rolls back down. Like Prometheus, we offer our traces to a system that feeds on us endlessly. As in the Panopticon, we live under an invisible gaze, but worse still: we have learnt to anticipate it, becoming our own jailers.

We are not only losing data; we are losing essential dimensions of being human: the inner life that lets us think without a witness, the forgetting that makes rebirth possible, the autonomy of being oneself, the heteronomy of being many.

Digital servitude needs no chains; it imposes itself through fluidity, seduction, habit. So the real question is no longer: “do I have something to hide?”, but: “how much longer will I be able to preserve what makes me a free being?”

Find the full transcript of this episode, along with all the sources and plenty more, at: The jailers of ourselves

Anyway, I’ve got nothing to hide… How many times have you said that while mechanically accepting a website’s cookies?

Yesterday morning, I watched my daughter check her phone. A simple, harmless gesture. And yet, in a few seconds, she had revealed her current mood, her sleep habits, her location, and even her plans for the evening.

Without knowing it, she was feeding her “invisible digital portrait”: a silhouette made of thousands of micro-traces that we scatter every day.

The problem? This portrait no longer belongs to you. It circulates, gets sold, grows richer. It can predict your desires before you even feel them. And in the wrong hands, it becomes a formidable weapon.

The real question is not “What are you hiding?” But “Why should you give up your privacy?”

In a world where forgetting becomes impossible, where every click sketches your future, protecting your data is no longer an individual luxury: it is the very condition of your freedom.

Find the full transcript of this episode, along with all the sources and plenty more, at: I have nothing to hide anyway…

What if the arrival of AI in medicine did not spell the end of doctors, but the beginning of a new era of care?

Since Hippocrates, doctors have drawn their legitimacy from their knowledge. Yet for the first time in modern history, the doctor is no longer necessarily the one who knows most. AIs diagnose faster, see what the human eye cannot catch, and sometimes even write answers that patients find more reassuring than a professional’s.

So should we fear the disappearance of the doctor? Or rethink their place, their role, their added value in a world where expertise is shared between human and machine?

Find the full transcript of this episode, along with all the sources and plenty more, at: With AI, the end of doctors?

Last week I talked about a point that is often misunderstood: for AIs, truth does not exist.

Today, I take the reasoning a step further. Because there is an even more deeply rooted misconception: believing that an LLM is a knowledge base. It is not. A language model generates probable sequences of words, not verified facts. In other words, it recites fluently, but it never cites.

That is exactly what I develop in my new podcast: why this confusion persists, and how to tell parametric memory from explicit memory, so that we can finally combine the two properly.

Find the full transcript of this episode, along with all the sources and plenty more, at: An LLM is not a knowledge base

An AI does not lie. Nor does it tell the truth. It does not know what is true or what is false; it calculates probabilities. Its “reasoning” boils down to guessing which word is most likely to follow the previous one, based on the billions of sentences it has been shown.

The result can be stunning: fluent, elegant, convincing. But this ease is only an illusion. What we read is not verified knowledge; it is a string of words that “fit nicely”. Sometimes right, sometimes wrong, sometimes neither.

So the real danger is not AI itself, but our human reflex: confusing coherence with truth. In other words, mistaking chalk for cheese. It is this subtle, almost imperceptible slide that opens the door to confusion, through ignorance of how it works and through excessive trust in whatever “sounds right”.

Find the full transcript of this episode, along with all the sources and plenty more, at: For AI, truth does not exist

I’m fed up with talk smothered in syrupy storytelling that turns a statistical engine into an Enlightenment philosopher. Fed up with these new AI “gurus”, freshly converted, who confuse producing fluent sentences with building a thought.

They would have us believe that an LLM meditates like an old sage beneath a bonsai, when all it does is churn billions of sentence fragments, like a cocktail shaker shaken at full speed. They dress it up with wisdom when it is merely a clever stochastic parrot, able to mimic intonation but incapable of grasping the deeper meaning of its own words.

SO, DEFINITELY NOT: AN LLM DOES NOT THINK.

And to go on claiming otherwise is to admit that you do not know how it really works.

Find the full transcript of this episode, along with all the sources and plenty more, at: GPT-5 has an IQ of 360… and my toaster meditates on Kant

Last week, I spoke to you about ants, those discreet beings who keep the world standing while others parade on stage. This week again, I won’t be talking about artificial intelligence, robots, algorithms or generative AI…

Today, I am staying in this very human, very intimate vein. Still about us. Always about us. Because before understanding what machines do to our thinking, perhaps we first need to understand what we have done to our own ability to think.

This time, I am taking you into subtler, more unsettling territory: our relationship with our own ideas. A silent slide that concerns us all, connected or not, technophiles or technophobes.

I promise, from next week, I will resume my series “AI in All Its States”. But for now, let me talk to you a little longer about this strange thing that happens to us when we stop inhabiting our own questions…

You type a question into your search engine. In 0.3 seconds, you have your answer. Satisfying, isn’t it?

And yet… something strange is happening. This disconcerting ease may be hiding a deeper transformation in our relationship with thinking.

There was a time when searching was already an act. When not knowing straight away was not a problem to be solved, but a space to inhabit. Today, we glide from one answer to the next, from one pre-chewed piece of content to another. We approve more than we choose. We apply more than we understand.

But what happens when thinking becomes optional? Between the seductive efficiency of our tools and our old habit of thinking for ourselves, a silent slide is under way. Not brutal, not visible. Just… comfortable.

The question is not whether technology is good or bad. It lies elsewhere, closer to home: do we still recognise our own voice when we think?

Find the full transcript of this episode, along with all the sources and plenty more, at: Does free will still exist in an intermediated society?