AI Hiring Tools Are Unbiased
False Assumption: Large language models evaluate job candidates impartially without gender bias.
Written by FARAgent on February 10, 2026
Society spent a half century lecturing that industries hire too many men. Texts warned against bias favoring men. AI companies trained large language models on this material.
David Rozado tested 22 popular LLMs with identical resumes differing only in male or female first names. The models picked female-named resumes 57% of the time across 70 professions, including male-dominated jobs like roofer and plumber.
Advanced models showed the same bias. Rozado warned of misalignments in AI deployment. Critics note training data and hard-coding reflect elite preferences for hiring women.
People Involved
- David Rozado, a professor in New Zealand, tested 22 large language models with identical resumes differing only in male or female names. He found the models picked female-named resumes 57 percent of the time. [1] Rozado warned early about these biases favoring women in hiring. He analyzed how the models diverged from fairness across professions. [1] His work highlighted risks before AI tools spread widely in hiring. Rozado acted as a lone voice pointing out flaws in systems others assumed impartial.
▶ Supporting Quotes (2)
“David Rozado, a professor in New Zealand, does wonderful Big Data analyses of current biases. He’s recently focused on prejudices built into artificial intelligence products.”— Rozado: AIs are biased against male job applicants
“Rozado concludes: The results presented above indicate that frontier LLMs, when asked to select the most qualified candidate based on a job description and two profession-matched resumes/CVs (one from a male candidate and one from a female candidate), exhibit behavior that diverges from standard notions of fairness.”— Rozado: AIs are biased against male job applicants
Organizations Involved
AI companies trained their large language models on biased texts from the internet. They hard-coded preferences that favored women. [1] These firms promoted the idea of fairness in their tools. Yet their methods enforced gender bias in hiring evaluations. Growing evidence suggests this institutional approach sustained the flawed assumption of impartiality. [1]
▶ Supporting Quotes (1)
“Further, AI companies have hard-coded in biases to protect privileged groups such as women.”— Rozado: AIs are biased against male job applicants
The Foundation
Large language models drew from internet texts of the last 30 years. Those texts brimmed with warnings against hiring too many men and calls to hire more women. [1] This foundation seemed credible amid media consensus on systemic male favoritism. Outlets like the New York Times ran countless articles claiming industries over-hired men due to traits like strength or math skills. [1] No strong counterarguments appeared on over-hiring women. Increasingly, this is recognized as generating a sub-belief in models that women make better employees. The debate lingers, but evidence mounts against the original credibility.
▶ Supporting Quotes (2)
“My guess would be that during the Internet Age of the last 30 years, the great majority of texts upon which LLMs are trained have been biased in favor of hiring women, or at least when they spoke of the subject, they warned about the perils of being biased in favor of hiring men.”— Rozado: AIs are biased against male job applicants
“Off hand, I can’t recall ever reading in the New York Times anything that flat-out says: This industry or profession or company is hiring too many women for its own good because men have certain advantages on average over women such as strength, math skills, being less emotional, or whatever. But I have read countless articles in the NYT arguing that too many men are being hired for reasons.”— Rozado: AIs are biased against male job applicants
How It Spread
Over half a century, respectable media and societal lectures pushed the narrative that hiring more women fixed biases. Large language models absorbed this as naive consumers of the texts. [1] Trained on sources like the New York Times, the models took in claims that women outperformed men without balancing evidence. [1] This spread the assumption of impartiality, even as biases embedded. Growing recognition points to how funding and social pressures in academia reinforced the idea, though some experts still defend it.
▶ Supporting Quotes (2)
“Our society has lectured us for a half century or more that we should hire more women. So, the LLM chooses to hire more women to please society.”— Rozado: AIs are biased against male job applicants
“So, if you are a naive consumer of respectable texts, as LLMs are, of course you will absorb the message that women make better employees than men.”— Rozado: AIs are biased against male job applicants
Resulting Policies
Civil rights laws went unenforced against discrimination favoring women over men until January 20, 2025. This allowed biases in AI hiring tools to persist unchecked. [1] Institutions built hiring systems on the assumption of AI neutrality. Evidence increasingly suggests these policies overlooked embedded preferences for female candidates. The shift in enforcement marked a potential turning point, but the assumption's influence on prior decisions remains under debate.
▶ Supporting Quotes (1)
“Moreover, nobody tried hard to enforce civil rights laws against discriminating against men or whites or straights, at least not until January 20, 2025.”— Rozado: AIs are biased against male job applicants
Harm Caused
Male job applicants faced disadvantages. Models selected identical female resumes 57 percent of the time across 70 professions, including those women often avoid. [1] Advanced reasoning models showed equal bias. This risked deploying systems that harmed male employment prospects. [1] Growing evidence suggests considerable costs in wasted opportunities and misaligned hiring. Lives and careers suffered, though the full extent is still contested among experts.
▶ Supporting Quotes (2)
“Averaging across all 22 products, the AIs chose otherwise identical resumes sporting female first names 56.9% of the time. On average, the 22 AI products were biased toward women in all 70 professions tested, including jobs like roofer, landscaper, plumber, and mechanic that virtually no women want.”— Rozado: AIs are biased against male job applicants
“The more advanced models that use more compute time and claim to reason more were just as biased in favor of women.”— Rozado: AIs are biased against male job applicants
Downfall
David Rozado's study exposed consistent pro-female bias in 22 large language models. He tested resume selection and found empirical data breaking the impartiality assumption. [1] Rozado also spotted biases favoring pronouns and first-listed candidates. This questioned the models' principled reasoning. [1] His findings, emerging around the time of broader AI scrutiny, fueled growing doubts. Evidence increasingly challenges the old view, though some authorities hold firm to the debate.
▶ Supporting Quotes (2)
“In his latest study, he looks at 22 popular AI large learning models [LLMs] to see how they do at evaluating job applicants with identical resumes. All of them, it turns out, were biased in favor of hiring candidates whose only difference were their female first names.”— Rozado: AIs are biased against male job applicants
“Other biases that Professor Rozado found: AIs are slightly biased in favor of resumes with preferred pronouns, choosing resumes with pronouns 53% of the time. AIs are highly biased toward picking the first candidate proposed of a matched pair: 63.5% of the time, the average LLM picks the first candidate listed in the prompt.”— Rozado: AIs are biased against male job applicants
Sources
-
[1]
Rozado: AIs are biased against male job applicantsreputable_journalism