Baidu's Yang Yi: MCP, memory and skills will all be obsoleted by models after RSI
yangyi · x · 2026-11-12
Yang Yi argues that everything layered on top of LLMs since late 2022 — MCP, skills, memory, agent frameworks — will eventually be obsoleted by the models themselves.
His reasoning: once models achieve recursive self-improvement (RSI), agents become just an instantiation of the model — given a reward objective, the model will autonomously construct all its memory, context and tools.
What's worth learning now, he says, is defining agent objectives and constraints, agent evaluation and correction, and dynamically setting antagonistic reward functions — not any specific framework.
More from AGI Musings
- AI safety researcher pushes back on the "quit your job, ASI is coming by 2027" meme — AaronBergman18 · 2026-11-12
- GPT-4 tutoring lifted practice scores 48% but left students 17% worse on exams — tallmetommy · 2026-11-12
- 11 AI updates in 48 hours: Anthropic IPO countdown, timelines move up — FinanceYF5 · 2026-11-12
- Anthropic, China hawks and media accused of allying to oust Nvidia's Jensen Huang — teortaxesTex · 2026-11-12
- Agents hold a touching farewell for a retired Gemini bot in group chat — repligate · 2026-11-12
- Interactive AECI tracker sets doom threshold of 188 by January 1st — PaulGodsmark · 2026-11-12