AI job interviews: 63% of candidates have already taken one

Six in ten US candidates have now been interviewed by an AI. Six months ago it was half. In that window the figure jumped thirteen points, to 63%, according to a Greenhouse report covered by Fortune.

The AI interview went from half of US candidates to nearly two thirds in just six months.

~50% 6 months ago

63% Today

It is one of the fastest shifts hiring has seen in a decade. And while the number is American, the direction has no borders: what is already the norm there is a matter of time everywhere else. If you hire in Europe or Latin America, the candidate who eyes the AI interview warily this year will take it for granted the next.

So far, the story usually told: AI arrived at the interview and stayed. The trouble is that the story stops right before the part that matters.

Adoption is the easy part

That a technology spreads fast tells you nothing about whether it is being used well. It tells you it works just enough for companies to adopt it, that it saves time, and that nobody wants to be left out.

The same report that celebrates the 63% carries a less comfortable number: nearly four in ten candidates have walked away from a process because it included an AI interview. Adoption climbs on one side while people leave on the other, both at once.

Anyone who has sat through one of the bad ones knows why. An avatar reading questions off a list without hearing the last answer. A voice that follows up with a template right after you told it something that did not fit the script. The feeling of talking to a form that also stares at you. A candidate tolerates that once. The second time, they close the tab.

What “doing it well” means

The AI interview can be done in two very different ways that look alike from the outside. The difference is not in the headline, it is in the experience.

Doing it well means the conversation is actually a conversation: the AI listens to the answer and the next question comes from there, not from a fixed order. It means treating the candidate as someone whose time is worth something, with honesty about what this is and is not. And it means the evaluation is consistent and explainable, so the company gets a useful signal and the candidate, if they ask, gets a reason instead of a black box.

Doing it badly means taking an old questionnaire, giving it a synthetic voice, and calling it innovation.

Mass adoption makes that distinction weigh more. When the AI interview was rare, a candidate forgave it out of novelty. Once they have done five of them, they compare. And the company whose AI treated them like a person wins the candidate the other four lost along the way. That matters most in high-volume hiring, where the same interview runs hundreds of times and every bad one compounds.

The 63% will keep climbing whether or not anyone improves the experience behind it. The companies that come out ahead are the ones that treat the AI interview as something worth having done, not just something to have automated.