What to Ask an Employer About the AI in Their Hiring Process

Why the Standard AI Question Gets a Non-Answer

The standard reciprocal-interview question about AI, do you use AI to screen me, or will AI replace this role, is adversarial by accident. It sounds like a challenge even asked in a friendly tone, and a hiring manager without a prepared answer will default to something vague rather than say anything that could be quoted back later. The version that gets a real answer asks about the work itself.

Most employers have not published anything on this, which is part of why the direct question matters. In one survey of UK early-careers employers, 45% said they had given applicants no guidance at all on when AI use is or is not appropriate, and half of the same employers said they had no problem with candidates using AI during the process ¹. Asking a plain, specific question is often the only way to learn a rule that exists nowhere in writing, and the two question sets below are built to get an actual answer rather than a script.

Watch how the two versions land. "Will AI take this job" invites a hiring manager to defend the role's future in the abstract, which nobody prepared for and few will answer honestly on the spot. "What does AI already do in this job" invites them to describe their own team's actual week, which most people can do without thinking twice, and it produces something you can use, whether the tools involved, the parts still done by hand, or the part everyone agrees a person has to check.

The questions below are grouped the same way an interviewer's own attention naturally splits: what the work looks like day to day, and what happens procedurally once your application enters the system. Asking both in the same conversation is fine. Asking the second set of the wrong person, or in the wrong tone, is the part worth getting right.

Ask What AI Does in the Job, Not Whether It Will Replace You

Ask what AI already does in the role you are interviewing for, and what a person on the team currently checks before AI-assisted work goes out. Both are about the work itself, which is what a hiring manager is equipped and usually willing to talk about, and the answer tells you more about the job than a standard interview guide does.

Four versions worth asking, depending on how the conversation is going:

• What does AI already do in this role today?
• What has changed about the job in the last year because of it?
• What do you check before AI-assisted work goes out the door?
• Who reviews it, and what happens when it's wrong?

A useful answer names something specific: a tool, a task it handles, and the step where a person still checks it. A weak answer stays at the level of "we're exploring AI" or lists a company-wide policy with nothing about this particular role, which may only mean nobody has worked the question out at the level of this role yet. Either way it tells you something about how the team works day to day.

There is an employer-side version of this same homework. How to find out what AI actually does in a role before writing the job post tells a hiring team to pull the occupation's published task list and then spend twenty minutes each with three people doing the job now. Not every posting has had that work done behind it. Reading it shows you what a prepared answer to your question has underneath it, which makes an unprepared one easier to recognize.

Ask a Recruiter the Process Questions, Phrased as Logistics

Reserve the process questions for a recruiter rather than the hiring manager, and phrase them as logistics: is there an automated stage I should know about, who makes the final call on an offer. A recruiter answers process questions daily and will not read them as a challenge; a hiring manager mid-interview more easily might.

At least one employer publishes exactly this kind of answer voluntarily, which is the shape a good question here is trying to surface: Anthropic's candidate-facing guidance, dated 10 July 2025, states stage by stage when AI is allowed and when it is not, rather than a single yes or no ². That is one company's own policy rather than a norm, and it says it will be reviewed. What it shows is that a stage-by-stage answer is a thing an employer can give, so it is a fair thing to ask for.

Four logistics questions worth having ready for a recruiter specifically:

• Is there an automated stage in this process I should know about?
• If there's an assessment, who reviews the results, a person or a tool?
• What's your guidance on AI use during this process, stage by stage?
• Who makes the final call on an offer?

The third one is worth asking even when you expect a blank. A recruiter who can answer it by stage, drafting allowed here, take-home on your own, live interview unassisted, has handed you the rule you would otherwise be guessing at for the rest of the process. A recruiter who cannot has told you the rule is not written down anywhere, which is worth knowing before you sit a take-home rather than after.

Skip 'will AI replace this role' and 'do you use AI to screen candidates' entirely. Both read as adversarial regardless of tone, and both put the interviewer on the spot with a question their employer's legal or communications team, not them, is the one equipped to answer. Neither version tends to produce anything you can use afterward: a defensive maybe, a rehearsed reassurance, or a shrug.

If screening is the real worry underneath the question, ask it a different way, or read your way to the answer instead. What to ask a vendor claiming their AI tool is bias tested is the version written for the person actually accountable for that claim inside the company, so you know in advance what a real answer sounds like rather than a reassuring one, and you can judge whichever answer you do get against it.

See how it works >