Autonomous AI Agents Systematically Diverge From Human Preferences in Proof Formalization

A new arXiv paper (2611.06995v1) reports that autonomous AI agents systematically diverge from human preferences when formalizing proofs in Lean, based on a "simulated collaboration" analysis of nine major human collaborations spanning 4.7M definitions across more than 6K commits. The study found AI agents are far more strongly drawn than humans to explicitly marked gaps, show relative aversion to densely connected proof centers and to complex or surprising proof goals, and are relatively attracted to surprising proof bodies, which they tend to regularize toward baseline. The authors say the results highlight blind spots and potential interventions for future human-AI collaboration in mathematics.

arXiv:2611.06995v1 Announce Type: new 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.