Emotional Reactions To AI-Generated Images: A Pilot Study Using Neurophysiological Measures

The rise of generative AI (GAI), powered by deep learning models and large datasets, is impacting on several fields due to its ability to produce highly realistic synthetic data. However, understanding emotional responses to AI-generated content, particularly images, remains a critical, understudied, and emerging area of investigation. To address this knowledge gap, we conducted a pilot study with 33 participants, using neurophysiological measures (EEG and SC) and self-reports (SAM) to compare emotional responses to real and AI-generated images. In addition, the effect of the participants' awareness of the image origin (AI-generated, real, and undetermined) was investigated. Linear mixed models (LMMs) with by-subject and by-stimulus random intercepts were fitted. Despite no significant results were found for the SAM questionnaires, findings emerged from the neurophysiological measures. The awareness of the image origin (AI-generated cue) showed an expectation effect on both EEG and SC, significantly affecting both valence and arousal. Additionally, the image origin itself (AI-generated) increased arousal, probably due to the hyper-realism, complexity, and novelty of the stimuli. Our contribution emphasises the importance of further research in this area to better understand the interplay between AI and human emotions.