When I started this newsletter, my biggest fear was running dry, not being able to keep up with a weekly rhythm. At first, I tried to play it safe: “No pressure, just publish when you feel like it.” Bad idea. I quickly realized I actually needed that pressure. “Constraint fosters creativity,” as a dear friend of mine would say. Eventually, my brain got used to the Sunday appointment. And even when I feel stuck, an idea always seems to emerge from the most unexpected place. It actually aligns with Albert Einstein’s idea that his best intuitions came to him while walking.
This week, I don’t have a clear idea in mind, no real thread. So I felt like lingering on that specific moment when you write without direction, just for the sake of it. An article is often a finished product. And even when it shows a line of reasoning, it’s usually a polished one, rarely all the surrounding “noise” that actually feeds thought. So why not just share, as they come, the things that made me pause this week?
AI doesn’t replace jobs. At least, that’s what I used to think. Fourth version written, hundredth in my head… my reasoning kept evolving. At first, I thought AI only replaced tasks. Then well-defined tasks. Then another question emerged: should every task that can be automated actually be automated? If we automate, if we settle for “good enough”… don’t we risk eroding taste and standards? And then another idea imposed itself: to replace something, don’t you first need to understand it? And if that’s the case, are we really talking about replacing jobs… or rather dismantling them, piece by piece?
Now, after reading a text by Anastasia Stasenko, another doubt creeps in. What if AI ends up replacing jobs by breaking the skill ladder? By absorbing the “small tasks,” the ones that used to train juniors, it dries up their learning ground. We gain time, yes… but at the cost of those slow, repetitive, sometimes tedious steps that actually made people competent. In other words, we’re sawing off the rungs at the very moment we claim to be climbing faster.
The problem is more insidious than expected. It’s not a brutal disappearance, it’s erosion. An erosion of the conditions that made it possible to become senior in the first place. And what are we doing about it? Not much. Well… not exactly nothing. A consultant told me that in his firm, access to Claude Cowork licenses isn’t given to juniors, it’s reserved for seniors. That struck me. It echoes something I’ve said before: the people best equipped to use generative AI are often the ones who could do without it in the first place…
“Less is not more when it comes to shoes”
To be a good user of AI tools… should we massively expand philosophy education? The temptation is to say yes. After all, what Anastasia Stasenko describes is a shift in perspective. Reading entire systems, confronting contradictory frameworks, learning to hold an idea without simplifying it too quickly. All of this resembles what philosophy produces when practiced seriously. But maybe the point isn’t philosophy as a discipline, it’s the kind of effort it requires. What those years of reading build is the ability to navigate ambiguity, to spot structures before diving into details, to compare frameworks without confusing them. In other words, judgment. And that’s precisely what AI does not do. A good AI user isn’t the one who writes the best prompt. It’s the one who understands what they are asking. The one who can evaluate an answer, see its limits, understand where it applies and where it doesn’t. Without that, the tool amplifies errors as much as good intuitions. Knowing what deserves to exist becomes the only skill that doesn’t automate.
“Soul in the game” vs “Skin in the game”
The self cannot be stored. Memory systems in AI make it possible to retain context, to give continuity to interactions, to avoid starting from scratch each time. By accumulating what a user says, the agent begins to “know” them. As Margot Lor-Lhommet points out, much of current research still focuses on this: how to better store and retrieve information to improve recall. But a concept from narrative psychology remains largely absent: the self is not stored, it is told. Identity is not a database you update, but an active construction that reorganizes the past to create coherence. We don’t just add elements to a story, we reinterpret them as we change. We already know how to build systems that remember what someone said. We are far less capable of building systems that can follow who someone is becoming….
Will our ego save us? That’s the bet of philosopher Gabrielle Halpern. She argues that faced with the infinite patience of machines, our own limits will suddenly become obvious. We realize we no longer really know how to listen. It stings. We thought we were great at human relationships… and then, not so much. That small blow to our ego might wake us up. Interesting… but is it enough? I don’t think so. What’s happening feels closer to grief. Have you ever experienced deep grief? It rewires your brain. That’s the scale we’re talking about. AI doesn’t challenge us just to make us better. It confronts us with the finitude of a model. It violently collides with a society where profession defines identity. What the machine interprets as resistance or pride is actually inner turmoil. It’s the cry of those who must reinvent themselves while the ground is giving way beneath their feet. The transformation is forced. And far more painful than a simple moral lesson. It’s the necessary passage through emptiness before we can, maybe, begin to redefine what it means to be human. And to be alive.
Build a market or redefine it? This video (in french) highlights a nuance we tend to overlook. Right now, all eyes are on China, with its systemic, pragmatic, almost relentless approach to AI and robotics. It produces, it ships, it delivers. On the other side, the United States seems more blurred, almost excessive in its pursuit of a “big leap.” But as Evan Kervella points out, it’s not really the same battle. On one side, you build a market, patiently, by accumulating use cases. On the other, you place a bet. That a breakthrough could reconfigure everything. And that’s where things become clearer. If robotics is just an execution problem, then China’s lead is already decisive. But if a real breakthrough still lies ahead, then those who seem behind today might be the only ones playing the right game. And in tech, it’s often these bets, fragile in appearance, that end up taking it all.
That’s it for today! If you have thoughts to share, feel free to leave a comment. Have a great Sunday 🙂
MD
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.