The internet has decided that the interesting question about AI and job-hunting is whether using it is cheating. You can find the argument everywhere: one camp says AI is a shortcut for people who can’t be bothered, the other says refusing to use it is leaving a tool on the table. Both camps are arguing about the wrong thing, and the argument is loud enough that almost nobody is asking the question that actually matters.
The question isn’t AI or no AI. It’s whether the words are yours or borrowed. And once you put it that way, the whole debate reorganizes itself around a line that has nothing to do with which tool you used.
Two very different things wearing the same name
When people say “I used AI on my application,” they could mean one of two things that couldn’t be more different. One person fed a model their real projects, their actual results, the messy specifics of a decision they made under pressure — and asked it to help them find the clearest way to say what they already did. Another person gave a model a job description and asked it to invent a candidate who fits. Same sentence. Opposite acts.
The first person is using AI to remember. The second is using it to pretend. And the reason the “is it cheating?” framing is useless is that it can’t tell these two apart — it flattens a person organizing their own truth and a person manufacturing a fake one into the same accusation.
AI is dangerous when it invents a version of you. It’s genuinely useful when it retrieves the real one.
The tell was never the AI
The folklore says you’ll get caught — that a recruiter, or a detector, will spot the machine and throw you out. It’s a comforting story for the “never use it” camp, and it’s mostly false. Detection tools are unreliable enough that OpenAI withdrew its own, and researchers have shown they wrongly flag non-native English writers at high rates. Recruiters know this. They’re not running your cover letter through a scanner.
What actually gives a fabricated application away is not a robotic sentence. It’s genericness — the eerie smoothness of writing that could describe anyone, because it was built to fit a posting rather than recall a life. And there’s a more unsettling finding underneath it. When researchers ran a controlled interview study, candidates who leaned on AI scored higher on the content of their answers and lower on how honest they came across. The tell wasn’t clumsiness. It was a faint sense that the person couldn’t fully stand behind their own words.
That’s the real cost of pretending, and no detector is required to collect it. A story that was never yours collapses at the first honest follow-up question — “walk me through how you decided that” — because you weren’t there when the decision the AI described got made. You can fabricate a paragraph. You cannot fabricate the memory it’s supposed to come from.
What remembering actually looks like
Here’s the part the cheating debate misses entirely: most people genuinely need help remembering. Not inventing — remembering. You did the work, but the number has slipped, the turning point has gone vague, the reason behind a good decision now takes a minute to reconstruct. That isn’t a character flaw; it’s what happens to experience that hasn’t been spoken aloud in years. A model that asks you the right questions and helps you organize what surfaces is doing something honest and useful. The words come back yours because they were always yours.
This is the line ELOQ is built on, and it’s worth being precise about it. AI pointed at your own experience — helping you recall what you did and rehearse how you’d say it— is preparation. AI pointed at a job description, asked to generate a person, is fabrication. The tool is identical. What changes is whether it’s reaching for something real or making something up.
You can fabricate a paragraph. You cannot fabricate the memory it’s supposed to come from.
The honest way to use it
So the next time you sit down to use AI on your search, don’t ask whether it’s allowed. Ask what you’re pointing it at. If you’re handing it your real material and asking it to help you say it clearly, use it without guilt — that’s not a shortcut, it’s the work. Whether they can tell stops mattering, because the answer to “did you write this?” is yes: you lived it, and the machine just helped you find the words.
If you’re asking it to invent someone who fits — stop, because that person has to show up to the interview, and it won’t be you. AI can help you remember who you already are. It can’t help you pretend to be someone else, and you wouldn’t want the job that required it.