World models vs LLMs: why next-token prediction still lacks internal representations
TheTuringPost · x · 2026-09-11
Kseniase shares a conversation with labenz and myprak on the gap between LLMs and world models: LLMs predict the next token but don't build the kind of internal representation world models aim for — organizing information around what actually matters for what happens next and ignoring the rest. The discussion also covers AGI and open source; full video linked.
More from AGI Musings
- When tech fluency is universal, the key skills will be non-technical — round · 2026-09-11
- Business Insider Asked ChatGPT, Gemini, Claude and Grok How AI Could End Humanity — coinfanking · 2026-09-11
- New Essay Argues AI Superintelligence May Never Be Achievable — horaciogarza · 2026-09-11
- Ex-OpenAI/Anthropic Pretraining Researcher Quits, Says Labs Are 'Racing to Self-Improving Superintelligence' — sjgadler · 2026-09-11
- Want to stop ASI? Quitting, laws, and treaties all fail—unless you conquer the world first — bennash · 2026-09-11
- Caltech undergrads defend AI math contest Mathathon against open letter criticism — Singularitarian · 2026-09-11