TeachBench: A Syllabus-Grounded Framework for Evaluating Teaching Ability in Large Language Models
This work addresses the need for better evaluation of LLMs as teaching assistants, though it is incremental as it builds on existing benchmarks and focuses on a specific domain.
The paper tackled the problem of insufficient evaluation of teaching ability in large language models by proposing a syllabus-grounded framework that measures teaching capability through student performance improvement after multi-turn instruction, revealing substantial variation in effectiveness across models and domains, such as good performance in mathematics but challenges in physics and chemistry.
Large language models (LLMs) show promise as teaching assistants, yet their teaching capability remains insufficiently evaluated. Existing benchmarks mainly focus on problem-solving or problem-level guidance, leaving knowledge-centered teaching underexplored. We propose a syllabus-grounded evaluation framework that measures LLM teaching capability via student performance improvement after multi-turn instruction. By restricting teacher agents to structured knowledge points and example problems, the framework avoids information leakage and enables reuse of existing benchmarks. We instantiate the framework on Gaokao data across multiple subjects. Experiments reveal substantial variation in teaching effectiveness across models and domains: some models perform well in mathematics, while teaching remains challenging in physics and chemistry. We also find that incorporating example problems does not necessarily improve teaching, as models often shift toward example-specific error correction. Overall, our results highlight teaching ability as a distinct and measurable dimension of LLM behavior.