“I can usually tell within a couple of follow-ups. A candidate who actually built the thing can explain it like they're talking to a non-technical person; the ones who rehearsed with AI loop the same jargon in a different order and go quiet when I ask what went wrong.”
“I can usually tell inside 90 seconds when a candidate’s answers are ChatGPT-polished: every response is too structured, and when I ask them to explain it like I’m not technical, they just rearrange the same jargon. The ones I actually place stumble a little, use plain language, and say I don’t know, but here’s how I’d figure it out.”
“I care less about whether the code is AI-generated and more about whether they can defend it. At Cookd we processed 31k candidates, and the ones who failed were rarely the ones who used ChatGPT; they were the ones who pasted a solution and could not explain why a tradeoff went one way.”
“I don't have a specific story about catching AI-generated code in an interview, but from my side of the table I watch for the gap between polished output and shaky explanation. Ask why they chose that data structure or what happens when the input changes; the ones who copied and pasted can't defend it.”
“I don't run the technical screen, so I watch for the gap between what a candidate can explain and what they claim to have written. When someone produces polished code but can't walk back through the trade-offs or why they picked that structure, that's the signal I flag for the hiring manager.”
“I don't screen code directly, but I do watch for the gap between a polished take-home and a candidate who can't explain why they chose that approach. The strongest signal isn't the code itself; it's whether they can walk you through the trade-offs they made.”
“I can't reliably spot AI-generated code from a resume, and honestly, neither can most hiring managers. The signal I trust is live problem-solving: put the candidate in a collaborative coding conversation and watch whether they can explain trade-offs, debug under pressure, and reason about why they chose this approach over that one.”
“I had a client whose team banned headphones and typing during virtual interviews and watched eye movement to catch candidates reading from a script. That caught some people, but the real filter happened before they ever got on that call: I probe one layer deeper on any technology they claim, and someone who cannot explain a tool they say they use daily falls apart in thirty seconds.”