AI in Learning and Feedback

Designing AI that helps people learn, and helps them act on criticism

In learning and in feedback, an AI system can easily end up doing the work that the person needed to do themselves. We design tools that leave that effort with the learner, and study what instructors, students, and authors do with them.

Learning by explaining

Explaining something to someone else is a reliable way to understand it yourself, but it requires a listener. Does it matter who the learner thinks they are explaining to? We built conversational agents that play three pedagogical roles — Tutee, Peer, and Challenger — and compared them against a control in a between-subjects study with 96 participants learning an economics concept [Xu, Z., Zhang, J., Tang, A., and Lee, Y. (2026). Who You Explain To Matters: Learning by Explaining to Conversational Agents with Different Pedagogical Roles. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26).] .

What instructors do with LLMs

Instructors are the ones who have to turn a general-purpose model into a usable classroom activity. We interviewed nine instructors about eleven activities they had built themselves, and used reflexive thematic analysis to work out how those activities line up with constructivist learning theory [Aurelia, E., Yeo, S., Lui, M., Law, E., and Tang, A. (2026). Instructors' Strategies in Creating and Implementing Constructivist LLM-Based Learning Activities. In Proceedings of the 2025 International Conference on Human-Engaged Computing (ICHEC '25).] .

Receiving feedback

Feedback that is accurate can still be unusable to the person receiving it. We set out to redesign teaching evaluation reports so that instructors would act on them. We worked through a design space of four feedback strategies against four presentation formats, and built six AI-augmented reports from it [Shang, R., Mallari, K., Au Yeong, W., Yasuhara, K., Tang, A., and Hsieh, G. (2025). Rethinking Teaching Evaluation Reports: Designing AI-transformed Student Feedback for Instructor Engagement. In Proc. ACM Hum.-Comput. Interact., CSCW320.] . We also interviewed fourteen HCI authors about how they interpret peer reviews and coordinate revisions across a team [Au Yeung, C., Stark, J., Li, J., Chevalier, F., Park, J., Kim, Y., and Tang, A. (2026). Supporting Reviewing Reviews: How HCI Authors Handle Peer Reviews of Manuscripts. In Proceedings of the 2025 International Conference on Human-Engaged Computing (ICHEC '25).] .

Publications

Xu, Z., Zhang, J., Tang, A., and Lee, Y. (2026). Who You Explain To Matters: Learning by Explaining to Conversational Agents with Different Pedagogical Roles. In Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26).
Acceptance: 25.3% - 1702/6730. Honourable Mention - Top 5% of all submissions
Aurelia, E., Yeo, S., Lui, M., Law, E., and Tang, A. (2026). Instructors' Strategies in Creating and Implementing Constructivist LLM-Based Learning Activities. In Proceedings of the 2025 International Conference on Human-Engaged Computing (ICHEC '25).
Shang, R., Mallari, K., Au Yeong, W., Yasuhara, K., Tang, A., and Hsieh, G. (2025). Rethinking Teaching Evaluation Reports: Designing AI-transformed Student Feedback for Instructor Engagement. In Proc. ACM Hum.-Comput. Interact., CSCW320.
Au Yeung, C., Stark, J., Li, J., Chevalier, F., Park, J., Kim, Y., and Tang, A. (2026). Supporting Reviewing Reviews: How HCI Authors Handle Peer Reviews of Manuscripts. In Proceedings of the 2025 International Conference on Human-Engaged Computing (ICHEC '25).