McKinsey & Company is adding a new round to its famously tough hiring process. Some entry-level candidates applying to the consulting firm this fall will have to face an AI avatar playing a client and hold their own in a roughly 10-minute conversation, according to a report by The Wall Street Journal.The pilot is aimed at undergraduates applying for business analyst roles in some of McKinsey's North American offices, and hundreds of candidates are expected to take it. The conversation happens on a laptop inside McKinsey's offices, right after the usual case-study challenge. Partners or interview assessors sit in the room and watch. The AI does not score anyone, and the hiring decision stays with humans.
How McKinsey's AI client interview round works
Saleem Zafar, a partner in McKinsey's Dallas office who helped design the tool, explained it with an example.
Say a city hires McKinsey to decide where to build a new stadium. The AI avatar plays the mayor, who wants to hear the candidate's recommendations."The mayor is going to have some very important things that matter to her, and she's going to want to hear your recommendations," Zafar told the WSJ. As the candidate talks, she may raise fresh concerns or add more context.The avatar has been trained on different client personalities, including a "professorial" type who wants to understand the reasoning behind a recommendation.
It is not built to play an unhinged client or throw gotcha questions. Zafar says it covers about 80% of the situations consultants deal with day to day. Assessors watch whether the candidate picks up on what the client said and acts on it, or steamrolls ahead with a prepared answer.
Why McKinsey wants to get past rehearsed interview answers
Blair Ciesil, who co-leads global recruiting at McKinsey, said the format helps the firm read human qualities that don't depend on preparation.
Candidates often walk in with polished, rehearsed stories, she noted, which makes it hard for interviewers to see who the person really is.McKinsey also says the tool makes the process fairer. Some human interviewers play tougher clients than others in role-play exercises, while an AI keeps the persona the same for everyone.AI in hiring has a patchy record, though. Amazon scrapped an experimental AI recruiting tool in 2018 after it showed bias against women, and many employers remain wary of new rules on AI in hiring.For candidates, the bar was already high. McKinsey reportedly accepts between 0.5% and 1.5% of applicants. Now, landing one of those offers could depend on how well they listen to a mayor who doesn't exist. If the pilot works, the firm plans to extend the AI client round to other roles next year.