Getting rejected for using AI in an interview is a joke when the interviewer is doing it too

I just got a rejection email on a Tuesday that actually pissed me off. I've dealt with plenty of "no's" before, but this one was different. The feedback said my technical skills were strong, but my "reliance on AI tools" during the assessment didn't show the "independent problem-solving" they wanted.

The irony? The guy interviewing me was quietly using AI on the same call, just forty minutes earlier.

The "honest" mistake during live coding

It was a standard live-coding session. Camera on, screen shared. I had to parse some messy input, transform it, and return a structured result. It's the kind of work I do every single day.

I handled it the way I actually work: I mapped out the core logic myself and used the assistant to stub out the boilerplate. I even narrated it out loud, telling him, "I'll let the assistant scaffold this part so I can focus on where it'll actually break."

I thought being transparent about my workflow was the right move. Turns out, that was the mistake.

How I caught the interviewer using AI

Everything felt fine until he went quiet. Not a "thinking" silence, but that awkward, uncomfortable kind of quiet.

The giveaway happened during a follow-up question about time complexity. I saw his eyes do the flick—that half-second pause where you're reading a screen. The answer he gave was too clean and too perfectly structured to be something a human just says off the top of their head on a random Tuesday.

I've used these tools daily for two years. I know exactly what someone looks like when they're reading a generated response because I've been that person. He was literally reading an AI's output while marking me down for using one.

The hidden rule of modern interviews

The real takeaway here isn't just the hypocrisy. It's that the actual rule isn't "don't use AI"—it's "don't let us see you use AI."

They want you to perform a 2019 version of a developer. They want you to hide the tools that everyone in the room is already using. They didn't reject me because I couldn't solve the problem; they rejected me for being honest about my process.

Testing for a job that doesn't exist

They think they're testing whether a dev can solve a problem without assistance, but that's a legacy test. Nobody on their team ships code without AI. The interviewer himself couldn't even handle a follow-up question without it.

The only test that actually matters now is: give the dev the AI, then see if they can tell when the model is lying.

I actually ran a test a month ago where I let AI write 100% of my code for 30 days without typing a single line of application logic. I managed to ship a real product, but the only reason it worked was because I knew when to look at a confident, plausible diff and say no. That discernment is the only skill that actually matters now.

The experience they ignored

During the interview, I flagged a specific issue with the problem's framing: the write path needed to persist before acknowledging, otherwise a retry could cause double-counting. That's a specific "ack-before-persist" bug that I learned the hard way years ago.

No AI gives you that kind of intuition; you earn it through expensive mistakes. That insight was worth more than the entire interview, but they were too busy worrying about my "reliance" on a tool to notice.