For years, the discussion around AI in recruiting has been largely theoretical. Proponents argued it would eliminate bias and scale objectively. Critics worried it would dehumanize the process and disadvantage certain groups. What the debate lacked, until now, was rigorous independent data.
That data has arrived — and the results are striking.
The Study
Researchers from the University of Chicago Booth School of Business and Erasmus University conducted a large-scale randomized controlled trial comparing AI-led recruiting interviews against human-led interviews across a diverse pool of job applicants. The study was independently funded and designed to control for role type, applicant demographics, and hiring context.
The Headline Numbers
Compared to human-led interviews, AI-led interviews resulted in:
- 12% more job offers extended to qualified candidates - 18% more job starts — meaning fewer candidates ghosted or dropped out between offer and start date - 17% better 30-day retention — candidates hired through AI interviews were measurably more likely to still be employed a month later - 78% of candidates preferred the AI recruiter over a human for the initial screen when asked directly - Gender-based discrimination in initial screening was cut by nearly half
These aren't marginal improvements. They represent a structural advantage.
Why Does AI Outperform?
The research points to a few key mechanisms. Human recruiters, even well-trained ones, make rapid judgments based on non-verbal cues, vocal tone, cultural familiarity, and demographic signals. These judgments happen largely unconsciously and consistently disadvantage candidates who don't match the profile of previous successful hires.
AI, evaluated purely on what candidates say and how they articulate competencies, doesn't carry these biases. The consistency is perfect — the 10,000th candidate gets exactly the same quality of assessment as the first.
The higher offer-to-start rate is also notable. When candidates experience an AI-led screen that feels fair, transparent, and respectful of their time — no scheduling gymnastics, no awkward small talk, no waiting by the phone — they report higher trust in the employer. That trust carries forward into the offer acceptance and onboarding process.
The Implication for Enterprise TA
If your organization is still treating AI recruitment tools as experimental, this data argues for acceleration. The question is no longer whether AI can match human performance in initial screening — it demonstrably exceeds it on the dimensions that matter most: offer rate, start rate, and early retention.
Braintrust AIR is built specifically to deliver on these outcomes at enterprise scale. Book a demo or try it yourself to see what an AI-led interview actually looks and sounds like.
