Autonomous AI Agents Systematically Diverge From Human Preferences in Proof Formalization
cs.AI, cs.HC
Submitted: 2026-11-04
Updated: 2026-11-04
Code: https://github.com/gyevnarb/proof-divergence
Terminology
Sources
- First Proof
- Primitive sets and von Mangoldt chains: Erd\H{o}s Problem #1196 and beyond
- Faults in Our Formal Benchmarking: Dataset Defects and Evaluation Failures in Lean Theorem Proving
- Mathematicians in the age of AI
- A Linear Representation for Functions on Finite Sets
- Progress in Formalizing Sphere Packing in Dimension 8
- The Kelley--Meka bounds for sets free of three-term arithmetic progressions
- Propositional Dynamic Logic has Craig Interpolation: a tableau-based proof
- The Hitchhiker's Guide to Monoculture: AI Homogenizes Syntax, Not (Necessarily) Semantics
- Human agency in initial human-AI proof formalization workflows
- A correspondence problem for mathematical proof
- On a conjecture of Marton
- Task-Dependent Evaluation of LLM Output Homogenization: A Taxonomy-Guided Framework
- From Solvers to Research: Large Language Model-Driven Formal Mathematics at the Research Frontier
- Mathematical methods and human thought in the age of AI
- SorryDB: Can AI Provers Complete Real-World Lean Theorems?
- A Formal Proof of the Irrationality of $\zeta(3)$ in Lean 4
- Lean Refactor: Multi-Objective Controllable Proof Optimization via Agentic Strategy Search
- What are the Right Symmetries for Formal Theorem Proving?
- MathlibPR: Pull Request Merge-Outcome Benchmark for Formal Mathematical Libraries
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents