BEA 2025 Shared Task

Pedagogical Ability Assessment of AI-powered Tutors

Motivation

This shared task will aim to assess pedagogical capabilities of AI tutors powered by LLMs and will focus on evaluating the quality of tutor responses aimed at student’s mistake remediation within educational dialogues aligned with the key pedagogical aspects grounded in learning science principles. The task will consist of two tracks and five subtasks, and all findings and resources will be made publicly available to support future research in this domain.

Organizers

• Ekaterina Kochmar (MBZUAI)
• Kaushal Kumar Maurya (MBZUAI)
• Kseniia Petukhova (MBZUAI)
• KV Aditya Srivatsa (MBZUAI)
• Justin Vasselli (Nara Institute of Science and Technology)
• Anaïs Tack (KU Leuven)