Young Adults’ Perception of Real versus AI-Generated Social Media Images: The Role of Personality, Confidence, and Perceived Realism

Abstract

Recent advances in generative artificial intelligence have made it increasingly difficult to determine whether images shared on social media are photographs or generated ones. The aim of this study was to investigate if Albanian young adults can correctly distinguish between real and AI-generated images and whether confidence, perceived realism, personality, and prior AI experience were associated with detection accuracy.  Participants aged 18–30 completed a visual classification task involving eight images, equally divided between real and AI-generated content. For each image, participants were requested to classify images as either AI or real, rating their confidence from 0 to 10 and the level of realism on a seven point scale. Personality was assessed using the Albanian standardized version of the Big Five questionnaire. The sample for preliminary analysis consisted of 56 participants after exclusion of two participants who reported using external verification tools. Participants recognized real images much more effectively than AI-generated pictures (72.7% compared to 27.3%), t(53) = 10.98, p < .001. Besides, participants were confident (mean = 7.86/10) in their judgments, and there was no difference in levels of confidence in correct and incorrect responses. AI-generated pictures were perceived as slightly more realistic than real photographs (M = 5.83 versus M = 5.62), t(53) = −2.05, p = .045. No significant associations were found between detection accuracy and Big Five personality traits or previous experience with artificial intelligence. Although data are still being collected, these preliminary findings suggest a strong tendency to accept convincing visual content as real, even when judgments are incorrect. The gap between accuracy and confidence highlights need to strengthen visual and AI literacy for young social media users.