Study on the Verification of AI Hyperrealism Effects and Intra- and Inter-Group Differences from a Cross-Cultural Perspective

This study aims to examine whether AI-generated "hyper-realistic" faces show the same perceptual bias of being perceived as "more real than real" in non-white samples (Asian populations), and it focuses on the moderating role of racial factors (other-race effect) in authenticity judgments.We hypothesize a replication of the AI hyper-realism phenomenon in china, where overall discrimination accuracy will be limited and proximate to chance levels (H1). Crucially, performance will be modulated by the in-group/out-group status of the stimuli. Consistent with the Other-Race Effect, participants are expected to demonstrate superior discrimination sensitivity (d') (H2) and significantly faster reaction times (H3) for in-group faces compared to out-group faces, indicating facilitated cognitive processing. However, regarding metacognition, we predict a specific calibration deficit for the in-group: participants will exhibit an "overconfidence effect," maintaining high subjective confidence even when erroneously classifying in-group AI faces as real, whereas out-group judgments are expected to show a different calibration profile (H4).