June 25, 2025
AI models like GPT-3.5 Turbo tend to rank Black male applicants lowest in simulated hiring scenarios placing them at a career disadvantage.
Emerging studies have raised serious concerns about the fairness of artificial intelligence used in hiring. A new study reveals that specific large language models (LLMs) show a preference for women while penalizing Black men, even when job qualifications are identical.
According to research published in “Robustly Improving LLM Fairness in Realistic Settings via Interpretability,” advanced AI models, such as GPT-3.5 Turbo, tend to rank Black male applicants lowest in simulated hiring scenarios.
In contrast, white and Black female candidates were more likely to advance in the process. The study measured the scores AI assigned to applicants. Then the technology simulated an 80 out of 100 threshold for progressing to the next hiring stage. At this cutoff, Black women had a 1.7-point increased chance of moving forward, and white women saw a 1.4-point gain. However, Black men were 1.4 percentage points less likely to advance.
The report suggests that while these models may appear to promote gender diversity, they also risk deepening racial inequities, particularly for Black men. The disparities persisted despite all candidates’ credentials being equal.
A separate study by VoxDev also examined how names affected AI screening decisions. Names commonly associated with white individuals were selected 85% of the time, whereas names typically tied to Black applicants were advanced only 10% of the time. These results mirror decades-old patterns of hiring discrimination, now seemingly replicated and expanded by automated systems.
University of Washington researchers added that male-associated names were preferred over female names in more than half of the simulations.
Taken together, the findings emphasize the risk that AI hiring systems may mirror societal bias. AI hiring systems may reinforce and intensify the biases. As AI becomes more integrated into human resources, experts urge employers to carefully evaluate these tools to avoid replicating harmful patterns.
RELATED CONTENT: Tyler Perry Halts $800M Studio Expansion Due To Advanced Artificial Intelligence, Admits ‘Jobs Are Going To Be Lost’
