Paper warns against optimizing proxy utility functions in AI decision theory

This paper explores the potential harms of optimizing proxy utility functions, arguing that this practice can lead to problematic outcomes when applied within decision theory frameworks. The author suggests that maximizing a proxy metric may not always align with true objectives and can introduce unintended negative consequences. AI

IMPACT Highlights potential theoretical pitfalls in AI alignment and decision-making frameworks.

RANK_REASON The item is an academic paper discussing theoretical issues in AI decision theory. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv

• paper
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →