AI is 'not smart' so what's next in artificial intelligence?
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Yann LeCun left Meta in 2025 after a decade as chief AI scientist and founded Advanced Machine Intelligence Labs (AMI Labs) in Paris to develop AI beyond systems like ChatGPT, Claude, and Gemini.
- AMI Labs raised more than $1bn (£760m) in seed funding from investors including Nvidia and the fund managing Jeff Bezos's private wealth — one of Europe's largest seed rounds.
- LeCun argues LLMs "are not particularly smart" because they accumulate and regurgitate knowledge but lack underlying understanding, demonstrated by their inability to reason about something as simple as a pen falling over.
- AMI Labs is developing Joint Embedding Predictive Architecture (JEPA), which creates abstractions of the real world to assess outcomes of actions by filtering out irrelevant information.
- Oxford University professor Ingmar Posner has spent four years with his team on a "mechanistic world model" inspired by a 2018 paper from David Ha and Jurgen Schmidhuber.
- Google's Dreamer World Model variant demonstrated the approach's potential last year by learning to collect diamonds in Minecraft through imagined future scenarios.
- LeCun plans to refine JEPA this year and deploy it in industrial settings by next year, ultimately aiming for "general generic intelligence systems" applicable to nearly anything with minimal training.
Why it matters: The $1bn backing from Nvidia and Bezos for LeCun's post-LLM approach signals serious institutional doubt that scaling current AI alone reaches superintelligence. If World Models succeed, robotics — long stuck on household tasks like ironing — gets a viable path forward, redirecting R&D away from pure LLM scaling.
Ask SkimNews
