BEA 2025 Shared Task: Evaluating AI-Powered Tutors’ Pedagogical Ability
BEA 2025 Shared Task evaluates AI-powered tutors’ pedagogical ability in educational dialogues, with five tracks and a registration deadline of April 23, 2025.
The BEA 2025 Shared Task focuses on assessing the pedagogical ability of AI-powered tutors in educational dialogues. Recent advances in Large Language Models (LLMs) have enabled the development of conversational tutors that can generate human-sounding dialogues. However, evaluating the quality of these tutors’ responses remains a challenge.
This shared task includes five tracks, inviting participants to develop systems to evaluate various aspects of tutors’ responses, such as mistake identification, location, pedagogical guidance, actionability, and tutor identity.
Participating teams can register using the registration link and submit their systems by April 23, 2025. The shared task website provides more information on the tracks, important dates, and evaluation criteria.
- Development data release: March 12, 2025
- Test data release: April 9, 2025
- System submissions due: April 23, 2025
- Evaluation results: April 30, 2025
- System papers due: May 21, 2025
- Final camera-ready submissions: June 9, 2025
- BEA 2025 workshop: July 31 and August 1, 2025
For more information, visit the shared task website or contact the organizers at bea.sharedtask.2025@gmail.com.
Tags: BEA 2025, Shared Task, AI-powered tutors, pedagogical ability, educational dialogues, Large Language Models, LLMs