Does AI-Generated Content Reduce Cognitive Load? A Comparative Examination of Instructional Materials Using Eye Tracking, Galvanic Skin Response, and Usability Measures

This study compares GenAI-generated, theory-based multimedia, and traditional instructional materials in cognitive load, visual attention, physiological arousal, and perceived usability. Fifty undergraduates evaluated all three conditions in a within-subjects repeated-measures design. Fixation count (FC) and total fixation duration (TFD) were recorded using Tobii X2-60 eye tracking, physiological arousal using Shimmer3 GSR, and usability using the System Usability Scale. Linear mixed and generalized linear models revealed higher FC (M = 21.922) and GSR (M = 2.213 μS) for traditional materials than for GenAI and multimedia (p < .001; partial η² = .03–.05). TFD did not differ significantly (p = .079). Usability favored multimedia (M = 61.96), followed by GenAI (M = 60.68) and traditional materials (M = 53.52; p = .002; partial η² = .163). GenAI and multimedia showed comparable biometric patterns. Interpretation is limited by five-second slide exposures, one unedited GenAI tool, a single-institution sample, and absent learning-outcome measures.