AI is Quietly Hijacking Job Interviews. Don’t Let It Fool You.

Let’s set the scene: You’re in the middle of an interview. The candidate has been chatty, engaging, maybe even charming. Then, you ask about their experience with a specific technology or a real-world problem they’ve solved. Suddenly, their eyes shift. Their posture changes. Their speaking style flips from “casual coffee chat” to “AI-generated LinkedIn post.” They pause more. They sip water. A lot of water. They start sounding like a TED Talk.

Congratulations, you’re not just interviewing the candidate, you’re interviewing their AI assistant, too.

I'm seeing over 60% of candidates are now using AI in interviews. Sometimes it’s blatant, sometimes it’s subtle, but make no mistake: AI is sitting in on your hiring process, whispering answers through second screens and hidden scripts. If you’re not paying attention, you might just end up hiring ChatGPT instead of a real problem solver.

Why This Matters

Hiring is already a high-stakes game. You’re not just looking for someone who knows the right answers, you’re looking for someone who can apply that knowledge, think critically, and solve problems your team actually faces. AI can fake competence, but it can’t fake experience. And if we don’t get better at spotting AI-assisted responses, we risk bringing in people who look great on paper (or in text-to-speech) but fold the second they’re on their own.

How to Spot the AI Impostors

AI-enhanced candidates leave clues. They’re not foolproof, but once you start recognizing the signs, they become impossible to ignore:

The Human-to-AI Trance Shift

• When answering broad, personal questions (e.g., “Tell me about yourself”), they sound natural, maybe even relaxed.
• When answering technical or experience-based questions, they suddenly shift into robotic precision. Their speech cadence changes. Their eyes scan an unseen script. Their conversational flow turns into a well-structured blog post.

The "Buffering" Tactics

• Repeating your question (“Can you say that again?”) more than normal, it’s a stalling tactic for AI lag.
• Sudden, frequent sips of water or coffee, classic delay maneuver. One candidate I interviewed took 15 sips in 30 minutes. Hydration is important, but that’s Olympic-level consumption.

Too Good to Be True Answers

• Their experience suddenly sounds perfect. Every challenge was met with textbook solutions. No mistakes, no ambiguity, no human struggle, just a highlight reel.
• Their words sound eerily polished, like they’re narrating a YouTube tutorial rather than recounting real-world experience.

The Non-Listening Listener

At the end, they ask perfectly crafted but generic questions. If you answered with “peanut butter and jelly,” they’d just nod and move on to the next scripted inquiry. They’re reading, not conversing.

How to Thwart AI-Generated Candidates

AI is good. But it’s not that good. Here’s how you level the playing field:

Video is a Must

• Phone interviews? Too easy for AI-enhanced candidates. Always use video so you can read body language, eye movement, and other tells.

Watch Their Words, Body, and Eyes

• Are they suddenly scanning a screen when answering tough questions?
• Does their tone change from human to robotic?
• Are they too perfect?

Ask About Challenges, Not Just Experience

• AI is great at summarizing past experience based on a resume. It’s less great at real-world problem-solving.
• Ask: “Tell me about a major challenge you faced recently. What unexpected problems came up? How did you adjust?”
• Press for details. Vague, polished answers are a red flag.

Test for Recent Knowledge

• Most AI models aren’t up-to-date. Ask: “What’s a new release, feature, or tool that excited you in the last 30-60 days?”
• If they struggle or provide outdated info, they’re likely relying on AI, not personal experience.

Throw in AI-Hallucination Questions

• AI models sometimes confidently spew incorrect information. I keep a list of technical questions that AI consistently gets wrong.
• If the candidate confidently repeats false information, that’s a major tell.

My Favorite Test: The Real-World Stumper

• I love throwing in challenges that our team struggled with, things no LLM could have answered for us.
• Example: “We hit a wall recently where no AI, no Stack Overflow post, no docs could solve our issue. We had to write our own solution. Here’s the problem, how would you approach it?”
• AI won’t help them here. The only way to answer is with real problem-solving skills.

Ask for a Screen Share or Live Whiteboarding

• Have the candidate screen share to work through a coding scenario, whiteboard a solution, or transform data in a sandbox.
• Watch how they interact with the problem, how long they take, and whether they’re comfortable navigating challenges in real time.
• While AI can assist here too, this method tends to separate those who know from those who are just parroting AI-generated responses.

Final Thoughts: AI is Good, But Humans Are Better

Look, I get it. Candidates want to put their best foot forward. If AI can help refine their answers, they’ll use it. But there’s a huge difference between using AI as a study tool and outsourcing thinking to a chatbot.

Our job as hiring leaders isn’t just to find people who sound good, it’s to find people who are good. AI can’t troubleshoot a live issue at 2 AM. It can’t navigate office politics. It won’t have a gut instinct for when a solution feels off.

So stay sharp. Ask better questions. Watch for the tells. And the next time a candidate drinks an entire gallon of water mid-interview, you’ll know, you’re not just talking to them. You’re talking to their AI, too.