The one big thing

When an AI screens resumes, it picks the ones that were written by AI, even when the person is just as qualified. Human-written resumes get tossed.

Here is the situation. Lots of companies get way too many resumes to read by hand, so they let an AI do the first cut. That AI likes resumes that sound like AI wrote them. So if you write yours plainly by yourself, you can get filtered out before any person ever sees it.

67–82%
How often the AI picked an AI-written resume over an equal human one.
23–60%
Better odds of getting shortlisted if you used the same AI the company uses.
2,245
Real resumes tested across 24 jobs. Worst hit: accounting, sales, finance.
Two quick quotes

“LLMs, when used as evaluators, systematically prefer resumes they generated themselves over equivalent resumes written by humans.”

The study, arXiv:2509.00462

“Instead of AI tools being used to find the applicant’s true abilities, you’re gonna find applicants that the AI thinks sounds like itself.”

Prof. Emma Wiles, Boston University
Is that cheating?

No. You are not lying about what you have done. You are just saying it in the style the robot likes. That is closer to spell-check than to faking your experience. The only real cheating is making up jobs or skills you do not have.

Which AI should you use? Guess from the job post.

Companies usually use whatever AI they already pay for. The job post often hints at which one. Match that, and you get the same-AI boost.

Claude
Big, careful companies. Banks, law, consulting, anything that talks about AWS, security, or compliance.
ChatGPT
The safe default. Startups, biotech, health, and anyone on Microsoft or Azure. Use this if you are not sure.
Gemini
Google shops. If they mention Google Cloud, Workspace, BigQuery, or ads, they probably use Gemini.

Spend two minutes on the job post and the company site. Which cloud do they use? Which industry are they in? That is your best guess. If nothing stands out, use ChatGPT, since it is the most common.

So should I write it like a human?

Not for the first round. A robot reads it first, and the robot likes robot writing. Use AI to get past the filter, then be fully yourself once a real person is talking to you. Keep the writing true, and let AI polish it instead of replacing it.

One catch. The researchers think this bias is a bug, and they showed it can be cut in half with a simple fix. So matching the AI works today, but it may not last. Treat it as a smart move for now, not a forever rule.