A Study on the Impact of Generative AI-Powered Reading Companion on College Students' Reading Effectiveness

While providing personalized support, generative AI may create a tension between cognitive efficiency and depth of thought. Current research lacks mechanistic evidence based on multimodal physiological data. To address this, this study employed a mixed experimental design, combining eye-tracking with functional near-infrared spectroscopy. The results showed that the generative AI-powered reading companion group performed significantly better than the self-directed reading group across various measures, with lower cognitive load, but no difference was found in reading presence. These findings indicate that AI-powered reading companions can improve reading efficiency and comprehension depth, optimize cognitive resource allocation, and provide neural mechanism evidence for AI applications in education.