How to Talk About AI in a Job Interview

Interviewers in 2026 hold two opinions at once, and you need to survive both. They expect candidates to use AI, and they get suspicious when candidates seem to depend on it. That suspicion is measured, not imagined: a set of experiments with more than 4,400 people, published in the Proceedings of the National Academy of Sciences, found a consistent “social evaluation penalty” for AI use — people who leaned on it were rated less competent and less diligent, and managers were less inclined to hire them. The penalty had one clear exception: it faded when the evaluator could see the AI was genuinely useful, or used it themselves.

That’s not a contradiction. It’s a filter. What interviewers are probing for is whether you direct the tool or the tool directs you. And more and more, the first round isn’t run by a person at all: an AI handles the initial screen before any human meets you.

Lead with the outcome, not the tool

The weak version: “I’m proficient with ChatGPT, Claude, and Copilot.” This is the new “proficient in Microsoft Office.” It claims familiarity and demonstrates nothing.

The strong version names a task, a number, and a judgment call: “Our campaign reports took about three hours each. I built a prompt template that drafts the first version from our analytics export, and now they take 45 minutes, mostly editing. I still write the recommendations section myself because that’s the part clients actually pay attention to.”

Notice what that last sentence does. It volunteers a boundary. Saying what you don’t hand to AI is the single fastest way to defuse the reliance worry, because it shows you’ve thought about where the tool’s judgment ends and yours begins.

Have one failure story ready

Anyone who has used these tools seriously has watched them produce something confidently wrong. Mentioning it works in your favor: “It kept inventing plausible-sounding stats for the industry section, so now anything with a number gets checked against the source before it ships.” A candidate who has never seen AI fail either hasn’t used it much or doesn’t check its output. Interviewers know both types.

If they ask whether you used AI to prepare

Answer honestly and specifically. “I used it to generate likely interview questions for this role and practiced answers out loud” is a fine answer and reads as resourceful. Evasion reads as having something to hide, and some interviewers now ask precisely to see how you handle it.

The pattern across all of these is the same: AI in your story should appear as something you deployed at a specific task, measured, and overruled when needed. Resumes get screened for AI keywords, but people get hired for owning the result.