Study Finds Most People Overconfident in Spotting AI Faces
Researchers developed the AI Face Test, revealing that natural object recognition skill—not intelligence or AI experience—best predicts the ability to identify AI-generated faces.
- Yesterday, researchers at UNSW Sydney and the Australian National University reported most people believe they can spot AI-generated faces but that confidence is outdated, the British Journal of Psychology showed.
- Because earlier visual cues no longer apply, UNSW researcher Dr. James Dunn noted that modern face-generation systems produce unusually symmetrical, well-proportioned faces that hide artificiality.
- Misplaced confidence could leave individuals and organizations vulnerable to scams and fake profiles, as relying on visual judgment alone is no longer reliable in social media platforms and professional contexts.
- Identifying high object-recognizers may guide defenses and training-data strategies, as this stable skill linked to radiologists and pathologists is 'baked-in' and resists simple training.
16 Articles
16 Articles
TechXplore
unsw.edu.auPeople are overconfident about spotting AI faces, study finds
Most people believe they can spot AI-generated faces, but that confidence is out of date, research from UNSW Sydney and the Australian National University (ANU) has demonstrated. With AI-generated faces now almost impossible to distinguish from real ones, this misplaced confidence could make individuals and organizations more vulnerable to scammers, fraudsters and bad actors, the researchers warn.
Can you detect an AI generated face?
Most people get around 11 out of 20. In a research paper published in the British Journal of Psychology, researchers from UNSW and the ANU recruited 125 participants – including 36 people with exceptional face-recognition ability, known as super recognisers, and 89 control participants – to complete an online test in which they were shown a series of faces and asked to judge whether each image was real or AI-generated. Obvious visual flaws were …
Coverage Details
Bias Distribution
- 100% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
