Akademik Veri Yönetim Sistemi
Journal of Computational Social Science, cilt.8, sa.4, 2025 (ESCI)
This research investigates the impact of fake news photos generated by artificial intelligence (AI) compared to real news photos. In the experimental phase, 20 reporters with at least 5 years of experience from various national media organizations in Turkey evaluated the “news impact” of selected published news stories in their original form and with AI-generated photos. The results indicate that while the criterion “news photo (image)” ranked highest for real photos, it ranked lowest for AI-generated photos. For real news, experts prioritized “photograph/visual of the news” as the most important criterion, followed by the presence of expert opinion and public authority. In contrast, for AI-generated photos, the presence of expert opinion was the top criterion. Interestingly, experts did not question the authenticity of AI-generated photos but rated them as less impactful. The fuzzy Bonferroni mean aggregation operator to synthesise expert opinions. The F-WENSLO analysis demonstrated a difference in the magnitude of criteria only in the “Photo/image of the news,” indicating a distinction between real and AI-generated photos in this aspect. However, the F-CRADIS analysis revealed that the overall ranking of news items did not significantly differ between real and AI-generated photos, suggesting that AI-generated photos produce an impact close to real ones. This study, limited to experienced reporters in Turkey, highlights the potential of AI to mislead experts and underscores the need for robust verification methods in news media. In this context, the scope encompasses any aspect of human–computer interaction (HCI) within the field of information science. Through this sentence, the research not only summarizes the effects of news photos but also addresses a broader aspect of information science, thereby encapsulating the significant findings of the study.