Everyone has AI now. How do you see what a candidate can actually do?
Stop trying to tell whether a candidate used AI. Give them real work, let them use AI openly, and watch what they do with it: whether they check its answers, and whether they could do the job without being carried by it.
Why the old question no longer works
A few years ago you could ask, "Did this person write this?" Today a cover letter, a CV and a take-home answer can all be produced in seconds. Detection tools are unreliable and they create a fight with honest candidates. And the question itself is out of date: at work, people use AI every day. Your new hire will too.
So the useful question is a different one: what does this person do when AI is in the room?
What to look at instead
Four things show up in real work and are hard to fake:
- Role skills. Can they do the core of the job, with real material in front of them?
- AI collaboration. Do they use the assistant well: clear questions, good use of what comes back?
- Critical verification. Do they check the AI's answer before passing it on, or just forward it?
- Judgment under uncertainty. When the information is incomplete or conflicting, do they make a sensible call and say what they are unsure of?
None of these shows up on a CV, in a cover letter or in a quiz. All of them show up in about half an hour of working on a realistic task.
What changes in your process
| Before AI | Now | |
|---|---|---|
| CV and cover letter | Showed effort and writing | Easy for anyone to polish with a chatbot |
| Knowledge quiz | Showed what they knew | A chatbot can answer most of it |
| Take-home task | Showed what they could produce | You cannot tell who produced it |
| Interview answers | Showed how they talk about work | Rehearsed with AI beforehand |
| Work sample with AI allowed | Rare, too slow to grade | Shows how they work, AI included |
The pattern: anything the candidate can finish alone and unobserved, AI can finish for them. A task where the AI is present and visible, and where the interesting part is what the candidate does with it, is the one that still tells you something.
What the research says
Work samples have a strong record. A 2022 re-analysis of decades of selection research, by Sackett, Zhang, Berry and Lievens in the Journal of Applied Psychology, placed work sample tests among the five strongest predictors of job performance. We do not yet have a validity study of our own, and we do not claim one.
How WorkProbe does it
WorkProbe is a work-sample assessment. Paste in a job description and it generates a job simulation from it. Or bring your own scenario. The candidate does 20 to 30 minutes of real work: messy documents, simulated colleagues, and an AI assistant they are free to use.
The simulation includes moments where the assistant's answer is wrong or incomplete. You see who catches it and who passes it on. WorkProbe then reports what each candidate did, with the evidence behind it, across the four skills above. You get a ranked shortlist in plain-word ratings, never scores or percentages, and interview questions anchored to what each person actually did. WorkProbe ranks and records. A person makes the hiring decision.
A simple way to start
- Pick one role you are hiring for now.
- Replace your screening quiz or take-home with a short job simulation.
- Allow AI, and say so in the invitation.
- Interview the top few, starting from what they did.