The hiring process is the worst place to pretend AI doesn't exist

The views expressed here are solely my own. They do not represent the opinions, positions, or policies of any current or former employer, client, or affiliated organization.

There's a story going around LinkedIn right now about a candidate caught using an AI app during a technical interview. The app was listening to the interviewer's questions and feeding him answers in real time. His eyes kept drifting to the left. By the fourth question, the hiring manager already knew.

He didn't get the role.

The post frames it as a cautionary tale about AI-assisted cheating. And on the surface, it is. But the moment I read it, a different question surfaced: why is this a problem in the first place?

Not the ethics part. That's straightforward — he was hiding something, and that matters. But structurally, why are we designing interviews that penalize the use of tools that the same candidate will use every single day once hired?

Most technical interviews — and a lot of strategic ones too — are designed around a core assumption: that the candidate's raw, unassisted knowledge is the best predictor of their future performance.

That assumption made sense once. When knowledge was the scarce resource, the person who had the most of it was probably the most valuable. But that's not the world we're operating in anymore.

AI doesn't just assist knowledge work. For a large and growing category of roles — content, strategy, product, operations, analysis — it has fundamentally changed what the actual job looks like. The person doing the work isn't retrieving information from memory. They're reasoning, deciding, iterating, questioning outputs, and knowing when to trust the tool and when to push back on it.

That's a completely different skill set. And none of it shows up in a traditional interview format.

So what are we actually measuring? The candidate's ability to perform under artificial pressure, without the tools they'll actually use, in a format that doesn't resemble the work in any meaningful way. We're not assessing fit. We're assessing interview readiness. Those are not the same thing.

Here's what changes when you allow AI openly in an evaluation: the floor rises for everyone. Basic answers become table stakes. The person with shallow reasoning and good prompting skills will produce a decent output — but so will everyone else.

What separates the candidates in that environment is exactly what you actually want to hire for: the quality of the thinking behind the output.

The candidate who knows how to ask a sharper question, challenge the AI's first response, iterate toward something more specific, and know when the output isn't quite right — that person will stand out immediately. Not because they know more. Because they think better.

The candidate who pastes the first response and calls it done? That's visible too. More visible than it would be in a traditional interview, actually, where surface-level answers can be disguised with confidence and rehearsed delivery.

Here's something I've been thinking about: what if the hiring process for knowledge roles reflected how the actual work gets done?

Send the candidate a scenario. Give them a realistic time window — say, an hour. Tell them to work through it using whatever tools they normally use, including AI, and to share the chat log from that session as part of their submission.

Then sit down with the top candidates and walk through their own conversation. Ask them why they pushed the AI in a certain direction. Ask them what they rejected and why. Ask them what they'd do differently.

That walkthrough is where the real evaluation happens. Not because you're watching them perform under pressure — but because you're watching them explain their own reasoning in a context where they have complete recall. They have the record right in front of them. They can point to specific moments, explain specific decisions, and show you how they think.

For the recruiter, this format produces something no traditional interview can: a documented artifact of the candidate's process. And that artifact is analyzable. The pattern of questions asked, the moments of iteration, the directions pursued and abandoned — that tells you significantly more about how someone thinks than whether they can answer a question correctly on the spot.

There are roles where foundational knowledge is genuinely non-negotiable. Medicine. Structural engineering. Security systems. When someone doesn't understand the underlying principles, the AI-assisted output can cause real harm.

But those roles already have rigorous credentialing systems, licensing requirements, and supervised practical evaluations. A Zoom question-and-answer session was never the right evaluation format for them anyway.

For the rest — the vast and expanding landscape of knowledge work where AI is already embedded in day-to-day operations — the argument for hiding it from the hiring process is getting harder to make.

The candidate who got caught wasn't the problem. The format that made hiding necessary was.

You can keep designing interviews for a version of work that no longer exists. You'll keep filtering for candidates who are good at performing under artificial constraints, hiring them, and then watching them use AI on the job anyway — only now without any insight into whether they're using it well.

Or you can start evaluating for what actually matters: not what candidates know in isolation, but what they can reason toward, build, and decide when they have access to the same tools as everyone else.

The companies that figure this out first won't just hire better. They'll build a process that actually reflects what they value — and that's a rare thing to be able to say about a hiring process.

The conversation that led to this article started with a LinkedIn post and turned into something I couldn't stop thinking about. If you work in talent acquisition or are rethinking how your team evaluates candidates, I'd genuinely like to hear where you're landing on this.