AIJun 22

Against Proxy Optimization

arXiv:2606.235975.9
Predicted impact top 87% in AI · last 90 daysOriginality Incremental advance
AI Analysis

It challenges the foundational assumption in AI and decision theory that proxy optimization is safe, relevant to researchers designing reward functions.

The paper identifies conditions where maximizing a proxy utility function leads to harmful outcomes, arguing that this undermines standard decision theory applications.

I discuss conditions under which maximizing a proxy utility function is harmful and suggest this poses problems for applying decision theory.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes