x

AI ALIGNMENT FORUM

AF

Fitness-seeking AIs — AI Alignment Forum

Fitness-seeking AIs

May 07, 2026 by Alex Mallen

How do fitness-seeking AIs (like reward-seekers) behave? How do they differ from long-term power-seekers (e.g. schemers)? What risks do they pose?

44Fitness-Seekers: Generalizing the Reward-Seeking Threat Model

4mo

2

46The case for satiating cheaply-satisfied AI preferences

3mo

4

42Fail safe(r) at alignment by channeling reward-hacking into a "spillway" motivation

Anders Cairns Woodruff, Alex Mallen

1mo

0

33Will reward-seekers respond to distant incentives?

4mo

1

29Recursive forecasting: Eliciting long-term forecasts from myopic fitness-seekers

Jozdien, Alex Mallen

1mo

0

51Risk from fitness-seeking AIs: mechanisms and mitigations

1mo

0