INTELLIGENCE ARTIFICIELLE 2026 - 2028 : se préparer au vertige qui vient

Anticipating the future is a perilous and uncertain exercise. But for AI, there are tools that allow it to be done in a grounded way: neural scaling laws, the equivalent of Moore's Law for AI. In this talk: • Why Claude 4.5 and Gemini represent a real generational leap • What scaling laws predict for 2027-2030 • The energy bottleneck: why future compute centers will need the power of one (or several!) nuclear reactors • Advances (world models, energy-based models, robotics) that could accelerate everything • The global geostrategic race around General AI (AGI) AI is the most transformative phenomenon of our era. Understanding its trajectory is essential to not endure what's coming. 📚 For more on AI's future, usage, and impact, discover my books and graphic novel: https://www.flavienchervet.fr/livres 📩 Weekly newsletter: hypernewsletter.fr