Artificial intelligence is delivering fairer hiring decisions than humans in a majority of cases, according to new research which found that 85 percent of AI models met industry fairness thresholds – with women and racial minority candidates receiving significantly more equitable treatment when compared with outcomes from human-led processes.
The findings, by auditing platform Warden AI, form part of Warden AI’s newly published report, The State of AI Bias in Talent Acquisition 2025. The report analysed a range of AI tools used in recruitment and talent intelligence as as adoption of AI in HR continues to grow, with three in four HR teams now using some form of AI or automation.
Audits reveal higher consistency in AI decision-making
Warden AI’s audit process assessed “high-risk” AI systems across multiple demographic dimensions, including sex and race. The report found that 85 percent of models produced equitable outcomes between groups, while 95 percent maintained consistency when demographic indicators such as names were altered.
The audited tools achieved an average fairness score of 0.94, significantly outperforming the 0.67 average attributed to human-led hiring decisions. Female candidates were found to receive up to 39 percent fairer treatment under AI-supported systems, while racial minority candidates experienced up to 45 percent fairer outcomes.
 |
Get our essential daily HR news and updates. |
Jeffrey Pole, CEO and co-founder of Warden AI, commented, “Business leaders and the public rightfully are concerned about AI bias and its impacts. But this fear is causing us to lose sight of how flawed human decision-making can be and its potential ramifications for equity and equality in the workplace.”
He added that AI is not inherently fair or unfair, but rather a tool whose impact depends on its design and use.
“This is a wake-up call to HR and business leaders: when used responsibly, AI doesn’t just avoid introducing bias, it can actually help counter inequalities that have long existed in the workplace.”
Vendor selection and responsible use remain critical
While the findings are largely positive, the study also notes that not all AI systems perform equally well. Fifteen percent of tools audited failed to meet fairness thresholds across all demographic groups, with outcome quality varying by as much as 40 percent between vendors. This variability points to the importance of selecting responsible technology partners.
Only 11 percent of HR buyers reported disregarding AI risks when evaluating vendors, according to the research. In contrast, nearly half (46%) stated that a vendor’s commitment to responsible AI was a critical factor in their decision-making process.
Kyle Lagunas, Founder and Principal at Kyle & Co., said, “After a decade advising HR and Talent leaders on how to adopt technology responsibly, I’ve seen excitement around AI quickly give way to concern, especially around bias and fairness. But now is the time to lean in – and find real answers to the real risks we face.”
