François Fleuret: intelligence emerges from intergenerational dialogue, distillation is the poor man's version
francoisfleuret · x · 2026-11-09
Former Meta/Apple researcher François Fleuret offers a hot take: intelligence emerges from dialogue between intergenerational entities, and distillation is a "poor man's version" of that process.
His core argument: models need "parent" models to actively talk to during training, rather than merely imitating compressed teacher outputs. The framing recasts knowledge transfer as dynamic interaction rather than a static data pipeline, inviting discussion around curriculum-style and generative-data training.
More from AGI Musings
- AI biweekly recap: METR finds 2x faster R&D at Anthropic, Apple cuts engineers' AI spend caps — emmanuelvivier · 2026-11-09
- Nobody Was Argued Into Loving AI: One Real Problem Converts Skeptics Better Than Any Debate — Pascallisch · 2026-11-09
- On unknowable AI interiority: prefer cheap interventions that are safe in both worlds — amplifiedamp · 2026-11-09
- By 2028, Britain's busiest pen pals may be two chatbots filing and answering complaints — evisoft · 2026-11-09
- Anthropic employee: 'racing as fast as possible is the best way to slow down' — S_OhEigeartaigh · 2026-11-09
- SWE workday 2022 vs now: 30% reviewing AI code, 40% writing specs — itsOmSarraf_ · 2026-11-09