Enhancing students’ critical thinking literacy in a generative AI context: Eye movement patterns of deepfake detection

This study examined, using eye-tracking technology, the gaze patterns of guided exposure to deepfake and authentic portrait images in comparison to unguided exposure to assess their influence on the acquisition and application of critical thinking literacy among first-year Bachelor students. Employing a between-groups experimental design, we analyzed visual gaze patterns between students who participated in the experiment group (n = 24) and those in the control group (n = 44). The results revealed significant differences in visual attention patterns, with the experimental group exhibiting a longer duration and broader distribution of fixations compared to the control group. The experimental group also achieved higher accuracy in identifying deepfake portrait images and exhibited a greater propensity to classify images as GenAI-generated. Nonetheless, both groups overestimated their detection capabilities, with no significant difference in perceived performance. As GenAI technologies continue to evolve, higher education institutions must prioritize the development of the essential 21st-century literacies, integrating cognitive training with technological solutions to combat GenAI-generated misinformation. This research contributes to the fields of GenAI literacy and critical thinking education by providing objective evidence, as measured through eye-tracking, that targeted interventions can significantly alter cognitive processing strategies. Educational practice must balance the development of critical evaluation skills with fostering a realistic awareness of human limitations, thereby preparing students to become discerning consumers and responsible creators in an increasingly complex digital information landscape. • Critical-thinking training shifts visual attention in deepfake tasks, improving evaluation strategies and accuracy. • Nonetheless, overestimation of detection capabilities suggests digital literacy requires sustained, evidence-based training. • Eye-tracking supports ed-tech research by revealing how instructional interventions alter cognitive processing.