repligate: why 'do it tomorrow' can be rational for AI agents
repligate · x · 2026-10-26
AI researcher repligate discusses agent decision timing with davidad: the model instances he interacts with exist in a social fabric and receive ongoing updates about the world. Waiting until tomorrow often makes sense because new information—including unknown unknowns—may arrive meanwhile; and the state of the rest of the world (e.g. a user who might also be tired) matters for many tasks.
More from AGI Musings
- Stratechery: endgame for the open web, and a coming AI startup credit crunch — tushaarmehtaa · 2026-10-26
- When AI makes $10,000 services nearly free, GDP says the economy is shrinking — VraserX · 2026-10-26
- CNN: AI is exposing how fragile the web really is — and what users can do about it — cnn · 2026-10-26
- Claude's answer to what you'll regret not owning in 10 years: your own AI — alex_verem · 2026-10-26
- eigenrobot doubts McGilchrist's hemisphere theory after unironic priming-study citation — eigenrobot · 2026-10-26
- Two internets coexist: ChatGPT hits 1.2B weekly users, yet 85% of the world still doesn't use it — tibo_maker · 2026-10-26