In many countries, competitive recruitment is more fiction than fact. And in the public sector in particular, interviews are often the stage where merit quietly disappears and “soft-rigging” takes over. Everyone knows it happens, but changing it is easier said than done.
That’s one of the reasons I think a new paper by Brian Jabarian and Luca Henkel is significant. They conducted a randomized controlled trial with 70,000 job applicants in the Philippines. The applicants were randomly assigned to be interviewed either by humans or by AI voice agents. Humans made all the hiring decisions, but the AI handled the interviews.
The results are extremely interesting. Candidates interviewed by AI received 12 percent more job offers, were 18 percent more likely to actually start the job, and stayed longer. When given a choice, 78 percent of applicants preferred AI. The paper is also an interesting case where AI reduces human biases instead of compounding them, with reports of gender discrimination dropping by nearly half.
Why do these effects occur? According to the study, it is because the AI stuck to the script. It asked the same questions and covered the same topics without getting distracted or showing favoritism. The authors refer to this as “controlled variance.” There was less interference from the interviewer, leading to a clearer signal for the decision-maker.
For public sector hiring, this suggests that AI could help interviews function the way they are supposed to: everyone gets an equal chance. Whether these effects would be replicated in government remains an empirical question. But it is worth looking into, especially given how flawed the status quo is.