article Article Summary
Sep 11, 2024
Blog Image

Middle school students perceive GenAI-powered teachable agents as beneficial learning companions, though challenges in problem presentation and adaptability remain.

The main goal of the study was to explore middle school students' perceptions of the roles, benefits, and challenges of using a Generative AI-powered teachable agent in a mathematics classroom.

Methods: The research involved classroom observations, focus-group interviews, and open-ended surveys with 108 sixth-grade students. The study utilized a GenAI-powered teachable agent designed to simulate a less knowledgeable peer, encouraging students to teach it math concepts.

Key Findings:

  • Students viewed the GenAI-powered teachable agent as a learning companion, facilitator, and collaborative problem-solver.
  • The agent helped create a student-centered learning environment, fostering autonomy and enhancing cognitive, social, and emotional capacities.
  • Students reported improved subject-matter knowledge and a positive attitude toward AI.
  • Challenges included the agent's lack of problem-presentation skills, inability to accommodate different learning styles, and technical issues.

Implications: The findings suggest that GenAI-powered teachable agents can support the learning-by-teaching pedagogical strategy, offering personalized and engaging learning experiences. They highlight the potential for AI to enhance educational environments by fostering student autonomy and improving learning outcomes.Limitations: The study's limitations include its focus on a specific educational context (middle school math class in the U.S.) and the proprietary nature of the LLMs used, which may introduce biases.

Future Directions: Future research should explore the application of GenAI-powered teachable agents in diverse educational settings, utilize open-sourced LLMs to mitigate bias, and investigate the long-term impact on student learning and attitudes.

Title and Authors: "Students' Perceived Roles, Opportunities, and Challenges of a Generative AI-powered Teachable Agent: A Case of Middle School Math Class" by Yukyeong Song, Jinhee Kim, Zifeng Liu, Chenglu Li, and Wanli Xing.

Published On: August 26, 2024

Published By: Computers and Society

Related Link

Comments

Please log in to leave a comment.