VivaTech 2025 just ended. Being there, one cannot but acknowledge the incredible vitality of the tech ecosystem—and the vibrant network of startups already shaping tomorrow's technological revolutions. This year, a particular emphasis was placed on expanding Europe's sovereign compute infrastructure. Ten days earlier, I had also attended USI 2025, a very different kind of tech event, and I was struck by how, in their own ways, both events felt like two sides of the same coin.
This year's theme for USI was "the unmeasurable part of digital." While we keep scaling our computing capabilities, we may be rediscovering how deep—and still mysterious—human thought truly is.
While AI models battle for the top spot in performance benchmarks, and major players compete to build massive computing infrastructures, USI brought together a remarkably diverse audience of technologists and philosophers—thanks Christian Fauré for this. One of my key takeaways was Luc de Brabandere's striking perspective throughout the long history of computation, tracing how epistemology has alternated between leading and following the advances in computation.
Then, Pascal Chabot invited us to explore the notion of "meaning"—a notion that's becoming ever more prominent, and increasingly central to individual choices. According to him, each time we interact with a screen, we connect to a kind of "digital superego," an unprecedented layer that shapes not only how we access information, but also how we perceive ourselves, others, and the future. He introduced the concept of "digitoses"—modern digital disorders—such as digital fatigue, eco-anxiety, rivalry with AI, and the rise of "machinoid" behaviors, where humans begin to mimic the very tools they use. "First we shape our tools, thereafter our tools shape us," as someone once said. As a matter of fact, the more "humanoid" they become, the more "machinoid" we risk becoming.
For some, the age of superintelligence is just around the corner. Others reduce it to stochastic parroting, including the most recent reasoning models—see Apple's paper—even though the timely critiques published along these lines feel, to some, like La Fontaine's fox pretending the unreachable grapes weren't worth having.
As for me, I believe we need to firmly hold together a couple of things: the faster our AI models evolve, the more often we must pause to reflect deeply on what we're actually building. The more technology permeates our daily lives, the more we need to sharpen our philosophical and anthropological perspective. The more autonomy we give to machines, the more we must empower our teammates.
As digital practitioners, our role is not to choose between acceleration and reflection, but to make sure one never happens without the other. And to keep asking the essential question at the heart of our relationship with machines—the one Luc de Brabandere left us with:
Who is programming whom?
