AI-Generated White Faces Perceived as More Real Than Human Faces, Reinforcing Racial Biases

Thumbnail Image

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

Research led by Australian National University found AI-generated white faces are perceived as more realistic than real human faces, due to biased training data. This hyper-realism reinforces racial biases, misrepresents people of color, and raises risks of misinformation and identity theft through AI-generated images.[AI generated]

Why's our monitor labelling this an incident or hazard?

The event involves an AI system generating highly realistic images, with the research indicating that this capability can reinforce racial biases and contribute to misinformation or deception online. These outcomes constitute harm to communities and potential violations of rights due to biased AI outputs. Since the harm is realized in the form of social bias and misinformation risks, this qualifies as an AI Incident rather than a mere hazard or complementary information.[AI generated]
AI principles
FairnessPrivacy & data governanceRespect of human rightsTransparency & explainabilityRobustness & digital securitySafetyAccountabilityDemocracy & human autonomy
Industries
Media, social platforms, and marketingDigital security
Affected stakeholders
Other
Harm types
Human or fundamental rightsReputationalPsychologicalEconomic/PropertyPublic interest
AI system task:
Content generation

Articles about this incident or hazard