Are They a Great Job Candidate or Just Using AI? 5 Questions to Tell – Inc. Magazine

This post was originally published on this site.

Amazon may be trying to stay at the forefront of artificial intelligenceI as a company in general. But that doesn’t mean they want job candidates looking to work there leaning on AI during the interview process.

In new guidelines, Amazon claims GenAI tools give candidates an “unfair advantage” and instructs internal recruiters to tell candidates explicitly not to use them.

“Failure to adhere to these guidelines may result in disqualification from the recruitment process,” the guidelines say, according to Business Insider.

Two thirds of job candidates are using AI

Amazon is certainly not alone in trying to figure how to handle job candidates using AI during the interview process. The 2025 Market Trend Report from recruitment firm Career Group Companies, found that 65 percent of job seekers are using AI during the job search process.

How to tell if job candidates are using AI: follow-up questions

That’s the issue. What’s to be done about it? You could start by emulating Amazon and loudly and clearly tell job candidates not to use AI. But how can you be sure that candidates aren’t cheating on their interviews and assessments with AI anyway?

Amazon’s guidelines offer a list of basic tells for recruiters to look out for, like candidates typing during interviews or offering answers that appear to be read out rather than spontaneous. You probably don’t need an expert to flag those for you.

In a recent article for the MIT Sloan Management Review Navio Kwok, a leadership adviser at recruiting firm Russell Reynolds Associates, offers another way managers can suss out whether a candidate is leaning on AI: Ask more probing follow-up questions.

Job candidates can and probably will use AI to think through how to answer expected interview questions. What they can’t have AI do for them is explain in detail the context and reasoning around their particular key career moments and milestones.

“Strategically, interviewing in the age of AI also requires effective follow-up questions to uncover deeper indicators of genuine expertise,” writes Kwok.

What does that look like in practice? Kwok offers five potential questions to ask.

1. Walk me through the details.

A bunch of science shows that nudging people to get into the weeds of a situation is one of the best ways to sniff out dishonestly. That apparently includes unethical or untransparent use of AI to prepare for an interview.

Kwok recommends asking questions like, “Walk me through, in greater detail, what you did to achieve the outcome” or “If someone unfamiliar with your approach — someone on a different team or without your domain expertise, for example — had to replicate your process, how would you explain it to them?”

Getting into the details, he explains, can “help in evaluating the extent to which candidates possess procedural knowledge. Do they really know how to execute a task or process?”

2. Why did you do it that way?

AI can often provide a convincing description of a business decision and its outcome. More often it will struggle to express the exact causes of why one led to the other. That’s why Kwok recommends asking something like, “What were the underlying principles that guided your decisions?” or “Why do you think your actions led to the outcome you achieved?”

Your goal, according to Kwok, is to understand if candidates understand why things work the way they do, or if they are just blindly applying a set approach or process.

3. How would the outcome have changed in a different context?

Examples of this type of question include: “In what situations (teams, organizations, industries) would your approach not work as effectively?” or “How would the outcome change if you were working with a different colleague or client?”

Again, the goal here is to probe not if a particular approach or solution works, but why it works in some situations and not others. It’s a red flag if a candidate proposes a blanket solution for all contexts without taking context into account.

4. What else did you consider?

Kwok sums up this fourth type of AI-detecting follow-up question as those that probe “the road not taken.” You might ask a candidate “What other approaches did you consider? r or “Who else did you consult before deciding on your approach?”

AI tools will generally not explain the alternative answers it did not offer. But a competent candidate should be able to walk you through the options they considered and rejected. This is also a great opportunity to see how candidates reason their way through problems.

5. What would someone who disagreed with you say?

This final type of follow-up question “asks candidates to act as a defense attorney” who can anticipate objections to their approach and defend against them. Questions might include: “What are the strongest counterarguments to your approach, and how do you respond to them?” or “What evidence did you gather to ensure that your approach was correct?”

In real life nearly every course of action has drawbacks, trade-offs, and uncertainties. Job candidates who truly know what they’re talking about (without an AI feeding them answers) should be able to talk you through them.

Better interviews are a happy side effect

What strikes me about all Kwok’s AI-detecting questions, is that they’re not just good for detecting problematic use of AI during job interviews. They’re all also simply solid interview questions that can help hiring managers get a deeper sense of candidates’ reasoning and expertise in general.

That these follow-up questions can help you tell if a candidate is using AI for the interview is really just a sweet added bonus.

The opinions expressed here by Inc.com columnists are their own, not those of Inc.com.

The application deadline for Inc.’s Best Workplaces Awards is this Friday, March 7, at 11:59 p.m. PT. Apply Today.