ProofTeller: Exposing Recency Bias in LLM Reasoning and Its Side Effects on Communication.
ProofTeller: Exposing Recency Bias in LLM Reasoning and Its Side Effects on Communication.
Talk by Alisa Kovtunova
- Location: APB-2026
- Start: 9. July 2026 at 11:00 am
- End: 9. July 2026 at 12:00 pm
- Research group: Automata Theory
- Event series: Research Seminar Logic and AI
- iCal
Large language models (LLMs) are increasingly applied in domains that demand reliable and interpretable reasoning. While formal reasoning methods can generate correct proofs, these proofs are often inaccessible to non-expert users. This raises a natural question: Can LLMs, when given a proof, faithfully interpret its reasoning and communicate it clearly? Recently, the authors have introduced ProofTeller, a benchmark that evaluates this ability across three tasks: (1) identifying key proof steps, (2) summarizing the reasoning, and (3) explaining the result in concise natural language. The benchmark covers three domains: Biology, Drones, and Recipes, representing scientific, safety-critical, and everyday reasoning scenarios. The authors find a consistent near-conclusion bias: LLMs tend to focus on steps closest to the final proof conclusion rather than on the most informative ones. A targeted human study confirms that explanations based on such steps are rated less appropriate for end users. These findings indicate that even when reasoning is provided, current LLMs face challenges in communicating key information in a useful manner, highlighting the need for LLMs that can communicate important details reliably.
The main paper was published in 2025 in the proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics. The talk will be also presented at DL 2026 and XLoKR 2026 in Lisbon.
