François Fleuret: Even SOTA Models Just Combine Existing Literature, Not Real Research
francoisfleuret · x · 2026-08-10
François Fleuret (Turing Award winner, Meta FAIR scientist) tweets that even the most advanced models, when brainstorming on a truly new problem, merely bring and combine existing concepts from literature, rather than engaging in the real research process.
More from AGI Musings
- Remembering Marvin Minsky: Turing Award Winner and AI Founding Father — moenig · 2026-08-10
- Prove You're Human: A Psychological Horror Game Questioning AI Ethics — nordicinst · 2026-08-10
- Economist Chad Jones Publishes Paper on AI and Our Economic Future in JEP — davilagrau · 2026-08-10
- Joke: Human Brains Are Just 2B Parameter Models Prone to Hallucination — yangyi · 2026-08-10
- Why Do We Say 'Thank You' to LLMs? Paper Introduces Embodied Hijack Hypothesis — MacrinePhD · 2026-08-10
- AI-Generated 3D Worlds Show Early Promise, Could Approach GTA-Scale Ambition — Scobleizer · 2026-08-10