1
Against Proxy Optimization
直指AI优化核心痛点：追求代理目标可能让系统偏离真实意图，一篇值得从业者深读的批判性视角论文
arXiv:2606.23597v1 Announce Type: new Abstract: I discuss conditions under which maximizing a proxy utility function is harmful and suggest this poses…
2
Flaws in the LLM Automation Narrative
论文批判LLM自动化叙事的结构性缺陷，揭示技术光环下的隐藏风险与过度承诺。
arXiv:2606.11166v1 Announce Type: cross Abstract: Large Language Models (LLMs) are increasingly described as performing at the level of human experts …