自主 AI 智能体在定理证明形式化中系统性地偏离人类偏好
Autonomous AI Agents Systematically Diverge From Human Preferences in Proof Formalization
AI 导读
一项 arXiv 研究提出“模拟协作”方法，考察 AI 智能体与人类在 Lean 定理证明中的偏好差异。研究分析了 9 个主要人类协作项目，涵盖 4.7M 条定义和超过 6K 次提交，发现智能体比人类更受显式标记空缺的吸引，相对回避连接密集的证明核心和复杂或意外的证明目标，并倾向于把意外的证明主体规范化到基线水平。
正文 · 原文
Abstract:Artificial intelligence (AI) systems have ushered in a new age of proof abundance, but helping humans understand these proofs requires new attention to the differences in how AI agents and humans pursue a proof-making task. We present "simulated collaboration", a methodology that allows us to examine two key questions: (1) the relative attention AI agents pay to surface syntax or, conversely, deeper underlying semantics, and (2) how and where AI agent preferences for what to do next diverge from human preferences. On a set of nine major human collaborations in Lean with 4.7M definitions across more than 6K commits, we find that, over the course of proof construction, AI agents (1) are far more strongly drawn than humans to explicitly marked gaps; (2) show a relative aversion to densely connected proof centers and to complex or surprising proof goals; and (3) are relatively attracted to surprising proof bodies, which they tend to regularize toward baseline. Our results highlight important blind spots and potential interventions for future human-AI collaboration in mathematics.
| Subjects: | Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC) |
| Cite as: | arXiv:2611.06995 [cs.AI] |
| (or arXiv:2611.06995v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2611.06995 arXiv-issued DOI via DataCite |
Submission history
From: Bálint Gyevnár [view email]
[v1]
Wed, 4 Nov 2026 18:34:25 UTC (388 KB)
来源：arXiv cs.AI · arxiv.org