Delegating Educational Tasks to LLMs: A Content Analysis of Evaluation Approaches
Résumé
LLMs have gone from theoretical computer science research to widespread use by the general public, with the emergence of increasingly powerful and popular LLMs models. These effective models have been incorporated into AIED research, but they raise several questions, since in this field, every prediction and every decision can have a significant impact on the learner. In this paper, we propose a study of LLMs' use in education, focusing on the evolution of the presence of these systems, their evaluation, and the tasks assigned to them. To this end, we conducted a meta-analysis, leading to the construction of a codebook based on the articles published from 2023 to 2025 editions of the AIED conference, ensuring agreement between annotators on a sample.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| licence |
