New EEG study reveals behavioral differences across AI learning modes in teens

A new exploratory study from Kashika Khurana and Ally Liew investigates how three distinct AI interaction modes—Tutor, Collaborator, and Solver—affect high school students' brain activity and behavior. With 48 participants (24 male, 24 female, ages 14–18), the within-subject design recorded frontal EEG spectral activity and quantified behavioral metrics across two counterbalanced quizzes. Statistical tests including Friedman tests, repeated-measures ANOVA, and paired t-tests revealed significant differences in all three behavioral measures (Initiation, Processing, and Stress) between the modes. The Tutor mode showed higher initiation, Collaborator mode balanced processing and stress, and Solver mode reduced proactive engagement.

The EEG data did not reach statistical significance, but descriptive patterns suggested neural activity differences warranting further investigation. Notably, the study explored short-term carryover effects of AI exposure, hinting that problem-solving behavior may shift even after the AI interaction ends. This work moves beyond the simplistic AI vs. no-AI dichotomy, offering a replicable framework for future human-AI interaction studies. For educators and edtech developers, the findings emphasize that not all AI-assisted learning is created equal—the mode of interaction can shape cognitive engagement and stress levels.