Strategy Executability in Mathematical Reasoning: Leveraging Human-Model Differences for Effective Guidance

Researchers identified why AI mathematical reasoning guidance is inconsistent and developed Selective Strategy Retrieval (SSR), a framework that improves AI math performance by combining human and model strategies. The method showed significant improvements of up to 13 points on mathematical benchmarks by addressing the gap between strategy usage and executability.

• →AI mathematical reasoning guidance fails due to a gap between strategy usage in solutions and strategy executability when applied to target models.
• →Human-written and model-generated mathematical strategies have complementary strengths that can be systematically combined.
• →Selective Strategy Retrieval (SSR) framework improves AI math reasoning by selectively retrieving strategies from multiple sources.
• →The approach achieved up to 13-point accuracy improvements on AIME25 and 5-point improvements on Apex benchmarks.
• →The research provides open-source code and benchmarks for reproducible mathematical reasoning improvements.