McKinsey Is Adding a Ten-Minute AI Client to Some Job Interviews

October 15 — Want a job at McKinsey? You may first have to hold your own against an artificial intelligence stand-in.

This autumn the top consulting firm is changing the interview process for candidates for some junior consulting roles, adding a client simulation. The exercise is meant to test how candidates think on their feet and how well they communicate with McKinsey clients — some of whom are more probing and harder to handle than others.

Although it is entirely a simulation, the roughly ten-minute AI conversation takes place in a real McKinsey office, with a partner or interviewer watching the candidate’s performance in person.

McKinsey partners say the AI tool does not score or evaluate candidates by itself. The avatar has learned several client personality types — for example, the academic type who wants to understand the firm’s recommendations in depth.

Saleem Zafar, a partner in McKinsey’s Dallas office who helped design the pilot, says the tool is not meant to simulate a client who loses their temper or asks deliberately loaded questions; the goal is to reproduce the situations that make up about 80% of a McKinsey consultant’s day-to-day work.

“What I’m watching closely is: did you pick up the key information in what the other person said? Did you respond to it, or did you simply ignore it? Did you stick to your own prepared thinking and force your answer on them?” Zafar said. “What we want to see is how you react when the information keeps changing.”

This autumn McKinsey is testing the virtual-adversary interview with undergraduates applying for business analyst roles in some North American offices, with several hundred candidates expected to take part. If the pilot goes well, the firm plans to extend it to other roles next year.

Companies across many industries are rethinking how they hire in order to adapt to the AI era and make better use of the technology. Some employers have added AI screening tools, and others have brought cheating-detection software into their hiring processes.

Landing an offer from McKinsey has long required getting through several rounds of demanding assessment. Candidates typically interview with partners and other interviewers, then take part in a case challenge in which they solve a problem on the spot.

Now, after the case challenge, candidates also hold a conversation with an AI avatar on a laptop.

“This way we can assess candidates’ overall interpersonal skills, without the effect of advance preparation,” said Blair Ciesil, a McKinsey co-leader for global recruiting. “Some candidates prepare a well-rehearsed presentation. It is not that those experiences are not impressive, but they make it hard for an interviewer to see the candidate’s genuine underlying qualities.”

Zafar gave a hypothetical example: a city hires McKinsey to study where to put a new sports stadium. In that scenario, the AI avatar plays the mayor, questioning the consultants’ proposal.

“The mayor will have some priorities she cares about deeply and will want to hear your recommendations. But as you explain your proposal, she may raise more concerns or add more background information,” he said.

Plenty of companies are reluctant to use AI in hiring, however, fearing bias or running foul of fast-changing AI rules for recruitment. McKinsey partners say the tool is intended to make hiring fairer, because different human interviewers vary in how tough they are during role-play exercises.

Zafar says hiring requires predicting whether “this person will be able to do the job in the future.” The new tool, he says, is meant to help interviewers see more clearly “what this person is really like — what their genuine side is.”