The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
The AI tool Face Depixelizer, designed to reconstruct high-resolution faces from pixelated images, repeatedly transformed images of Black individuals, including Barack Obama, into white faces. This incident highlights significant algorithmic racial bias stemming from the tool's training data, raising concerns about misrepresentation and discrimination in AI systems.[AI generated]
Why's our monitor labelling this an incident or hazard?
The AI system is explicitly mentioned and its use has directly led to a form of harm—algorithmic racial bias in image reconstruction. This bias can cause harm to communities by perpetuating racial misrepresentation and discrimination. Since the harm is occurring through the AI system's outputs, this qualifies as an AI Incident under the framework, specifically under harm to communities or violation of rights.[AI generated]
