<scp>AIming</scp> High: Do Goal Structures Matter in Learning With <scp>ChatGPT</scp> ?

ABSTRACT Recent meta‐analyses indicate that learning with ChatGPT improves academic performance but reveals substantial heterogeneity in effect sizes. The present study sheds light on one theoretically plausible moderator of the benefits of learning with ChatGPT: the goal structure of the learning task. For this purpose, in an experiment, university students ( N = 104) learned four social psychology concepts using ChatGPT, with the task either embedded in a mastery goal structure, emphasizing the goal of making individual progress, or in a performance goal structure, emphasizing the goal of outshining others. Results showed that the mastery goal structure fostered the acquisition of conceptual knowledge in comparison to the performance goal structure. Furthermore, the performance goal structure increased pressure, anxiety, and the request for non‐essential/peripheral details aimed at appearing smart in comparison to others, although these hardly contributed to learning. We conclude that goal structures can moderate the benefits of learning with ChatGPT.