ChatGPT for good? On opportunities and challenges of large language models for education

This position paper discusses the potential benefits and challenges of educational applications of large language models, viewed from both student and teacher perspectives. After briefly reviewing the current state of large language models and their applications, it highlights how these models can be used to create educational content, improve student engagement and interaction, and personalize learning experiences. The authors treat the technology as here to stay despite critical views and bans in some communities.

On the challenges side, the paper argues that using large language models in education requires teachers and learners to develop competencies and literacies to understand both the technology and its limitations and unexpected brittleness. It calls for clear strategies within educational systems and a pedagogical approach centered on critical thinking and fact checking, while noting risks such as output bias, the need for continuous human oversight, and potential misuse. The authors conclude that, handled sensibly, these challenges can themselves become opportunities to teach students about the biases and risks of AI, and offer recommendations for responsible and ethical use.