Researchers Debate Using Video Games to Probe Intrinsic Motivation and Goodhart
A debate erupted over jdpressman's proposal to use video games to probe intrinsic motivation, with critics arguing optimized games exploit accidental features of human psychology. Pressman laid out an alignment roadmap using a "Weave algorithm" and Gillen quantizer to tackle Goodhart effects, arguing alignment's basic problems are well known yet almost no one is solving them directly.
2026-10-16 ~ 2026-10-16 · 4 related posts
- Researcher: alignment's basic problems are known, yet almost nobody works on them directly — jd_pressman · 2026-10-16
- Debate: do gacha games reveal human values the way RLHF degenerate states reveal models? — enormochungus · 2026-10-16
- jd_pressman: alignment's basic problems are known — but nobody is solving Goodhart in RL — jd_pressman · 2026-10-16
- jd_pressman shares alignment roadmap: Weave algorithm plus Gillen quantilizer against Goodhart — jd_pressman · 2026-10-16