Does AI Beat Humans at Recruiting?

But the results favored AI. Candidates interviewed by a voice-AI agent rather than a human were 12 percent more likely to receive a job offer, 18 percent more likely to start the job, and 17 percent more likely to remain at work after 30 days. Of the candidates who actually accepted an offer and started the job, those screened by AI were 6 percent more likely to stay in their roles at least 30 days.

What’s more, most applicants—80 percent—chose voice AI over a real person for their interview. The researchers cite three reasons applicants may have preferred AI: the perception of fairness or objectivity, a more standardized interview process, and convenience. “The attractiveness of AI is that it’s always available,” Jabarian says. (For more, listen to “Are AI interviews better?” on the Chicago Booth Review Podcast.)

He points to an advantage to automation that came through in the research: AI interviewers elicited more interactive, engaged responses from candidates, while helping them steer clear of negative conversational habits such as vague answers. Participants who engaged with AI agents spoke more clearly and directly, used fewer filler words, and provided more relevant information than participants in human-led interviews.

It’s important to note that once these applicants were hired, they performed well when interacting with humans in customer service positions, Jabarian points out. In other words, AI didn’t give a boost to applicants with weak people skills. Rather, “the pattern is more consistent with improved signal extraction: AI changes how information is expressed in the interview, and that information predicts real match quality,” says Jabarian. “AI is better at picking up on linguistic features—like interactivity during a conversation—that actually matter for a job decision made by humans.”

He also notes that the AI interviewers appear to have avoided some of the biases that have plagued AI in other studies, according to a participant survey. “The reported gender discrimination is half for people who got job interviews with an AI than with a human,” Jabarian says.

While he cautions that voice AI, as it currently exists, is likely best suited for high-volume, structured roles, the technology’s rapid evolution suggests that it could soon play a part in many kinds of hiring. For now, human judgment remains essential, and most final decisions are made by recruiters using AI-generated transcripts rather than voice-AI interviews. But as automation expands, the research suggests that machine screening could evolve into AI selection.