AI Hiring Systems Found to Exhibit Greater Bias Than Humans

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.

Studies by researchers at Princeton and the University of Chicago reveal that large language models used in AI hiring systems not only inherit human biases but can also develop new, more pronounced stereotypes, leading to discriminatory hiring decisions against candidates based on gender, ethnicity, and age. This raises significant concerns about fairness in AI-driven recruitment.[AI generated]

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

The article explicitly involves AI systems (large language models) used in hiring decisions, demonstrating that their use leads to discriminatory stereotyping and biased hiring recommendations. This bias is not only inherited from training data but also emerges from the AI's own decision-making process, causing harm to groups of applicants by unfairly limiting their employment opportunities. Such outcomes constitute violations of labor rights and unfair employment practices, fitting the definition of an AI Incident. The harm is realized and directly linked to the AI systems' use in recruitment simulations, with implications for real-world hiring where such AI is increasingly deployed.[AI generated]
AI principles
FairnessRespect of human rights

Industries
Business processes and support services

Affected stakeholders
WomenGeneral public

Harm types
Human or fundamental rightsEconomic/Property

Business function:
Human resource management

AI system task:
Organisation/recommenders


Articles about this incident or hazard