The interview is the one part of hiring where two or more people actually meet. We’ve been quietly taking it out.

Before we do, it’s worth asking what the person on the other side was doing there.

1. More of hiring no longer needs a person

David Kraase asked a question in his newsletter, The Context Engine, that I haven’t been able to shake: Do people still need people?

AI lets one person do more alone. So what were we getting from other people, besides their help?

I keep asking the same thing about interviews.

AI can already screen CVs, book the call, run the first interview, transcribe it, summarise it and score it.

Candidates use it too. They practise with it, polish their answers with it, and sometimes bring it into the interview itself.

Some of that is good. It saves time, and time is the one thing nobody can get back. Kraase put it well: “We cannot recover Tuesday afternoon.”

And AI can genuinely help us choose. In one field experiment, candidates shortlisted with the help of an AI interview report passed the final interview 17.5 to 20 percentage points more often than those picked from their CV alone. The paper is called Better Together. Notice who ran that final interview: a person.

So the question isn’t whether AI belongs in hiring.

It’s what the person in the interview was doing that never made it into the job description.

2. What was the interviewer doing besides asking questions?

Most hiring teams have one. The interviewer who asks one more question when everyone else is ready to move on.

“Why this way, and not the simpler one?”

“What would you do if the requirement changed tomorrow?”

Sometimes you wish they’d stop, so the interview can finish on time. Then the hire starts, and you realise that one question showed you something nobody else saw.

That interviewer isn’t always right.

Experience can sharpen judgement, or trap you in an old answer. But they stand somewhere the candidate’s CV doesn’t reach. They know the team. They know what the last hire struggled with. They know what “good” looks like here, not in general.

That’s context. And we rarely count it when we count interview hours.

In my own interviews, a lot of the work never shows up in a score. I start with an ice-breaker, because a nervous candidate can’t show anything to me of what they know.

I notice what kind of help someone needs during the live challenge too. Be it a tiny question guiding them to think in a different direction. The name of a library they can use. Or sometimes, pointing out that stubborn typo, that one letter which sits out of place, which the nervous mind of the candidate completely filtered out of the picture.

And when an answer is wrong or half right, I lead them to the right one.

Most of that is not on the scorecard.

All of it changes what I see.

3. Couldn’t AI be that interviewer?

Partly, yes. AI can ask a follow-up. It can catch a gap in an answer. Sometimes it notices something a tired interviewer misses on their seventh call of the day. P.S. I don’t recommend you run 7 technical interviews in one day. 7 HR screening calls, maybe yes.

I also can’t defend every human interview by saying “only a person can ask a good question”. That isn’t true anymore.

But look at what happens before the first question.

Someone decided what this role needs. Someone wrote the questions, and decided what a great answer sounds like. Every AI interview starts from those human choices.

The difference is where we all start from.

I was a teaching assistant for Tim Roughgarden’s online algorithms course at Stanford. Algorithm questions are one of the things I know best. They’re also the thing AI now answers in seconds. If an interview is only “can you solve this”, you really don’t need me in it.

What a person brings is everything around the answer. Having done the job. Fixing what was broken in the middle of the night because the system was down. Knowing the team. Being able to tell the difference between someone who memorised a solution and someone who understands it to its core.

4. What is the candidate missing?

Let’s not forget. There’s another person in the interview.

A candidate doesn’t only answer questions. They’re deciding whether they want to work with you. The interviewer is often the first real person from the team they meet. How that person listens, what they ask and how they explain the work is the company, for that hour.

Anthropic recently looked at how AI changed work inside its own walls. Even there, where AI agents do most of the hours, the report lists “a sales call, an interview, cross-functional alignment” among the parts of work that remain human.

Candidates feel it too. In Greenhouse’s 2026 survey, the most common reason people gave for walking out of a hiring process was a pre-recorded video interview scored by AI, with no human there at all. In Europe, from December 2027, employers using AI like this will need a trained person overseeing it, and will have to tell candidates.

Take the person out, and the candidate loses the only chance to see who they’d actually be working with.

5. What do we want the time back for?

AI gives hiring teams real hours back. That’s worth having.

But it’s worth deciding what those hours are for. Here’s what I’d spend them on:

  • Better questions. Write them before the first candidate, and agree what a great answer sounds like.
  • The conversation. Let AI book the call and take the notes, so the interviewer can actually listen.
  • Feedback. Every candidate I interview gets written feedback within 24 hours. An AI could draft it. A person has to stand behind it.
  • The decision. Read the notes, compare the scores, talk it through. The part that affects someone’s career shouldn’t be the part we rush.

We count the hour saved by skipping an interview. We rarely count the months spent with the wrong hire, or the great candidate who left because nobody seemed to care.

TL;DR: AI can take a lot of work out of interviewing, and it should. But the interviewer was never just asking questions. Spend the time you save on the parts only a person can do.

What does the best interviewer on your team do that would never show up on a scorecard?

Want People in Your Interviews, Without the Hours?

That’s what we do at AssessDev. Our Tech Assessments are structured technical interviews run by a senior engineer or CTO, human to human, with a report within 24 hours. And if you need someone inside your team, our Embedded Tech Interviewer works with you on every role, from the job description to the decision.

Tech Assessments

Structured technical interviews run by a senior engineer or CTO, human to human, with a report within 24 hours.

Embedded Tech Interviewer

Someone inside your team on every role, from the job description to the decision.

Sources

This post first appeared in Shortlisted by Kat. Photo by laura adai on Unsplash.