No, an AI didn't reject your CV

You applied three weeks ago and heard nothing back, not even a rejection email. Silence like that demands an explanation, and the one everyone offers is the same: nobody read your CV. An algorithm scanned it, found a keyword missing and moved on. If that's true, then spending ten minutes on a careful application is wasted effort, and you might as well automate everything.

I read around 500 applications a week, in one of the most advanced ATS on the market. I'd like to walk you through what actually happens to your application after you hit send, because the reality is quite different from the story, and understanding it will change how you apply.

What candidates believe

Spend ten minutes on LinkedIn or Reddit and you'll find the same beliefs repeated everywhere:

• "Recruiters don't read CVs anymore." Everything is pre-filtered by AI, and humans only see what the machine allows through.
• "75% of CVs are rejected by the ATS before a human sees them." A number quoted everywhere and sourced nowhere. Nobody can point to the study behind it, because there is no serious study behind it.
• "You have to stuff your CV with keywords to pass the filter." Hence the CVs listing 70 skills in tiny font.
• "Writing a real answer is a waste of time." Since a robot reads it, let a robot write it.

These beliefs are understandable. When you've been applying for months into what feels like a void, a story that explains the silence is almost comforting. It's also a story with an industry behind it: ATS-optimized templates, resume scanners and auto-apply tools all sell better when the robot feels inevitable. The problem is that the story doesn't match how hiring actually works today, and believing it pushes candidates to apply in exactly the wrong way.

What an ATS actually does

An ATS is a database with a workflow on top. It stores applications, schedules interviews, sends emails and keeps a pipeline organized across dozens of open roles. Most of what it does is administrative.

It does eliminate some candidates automatically, in one narrow and well-defined case: binary criteria. If the posting says the role is based in Berlin and you answer that you're not there and won't relocate, you're out. The same goes if your salary expectation sits far outside the announced range, if you need a visa the company can't sponsor, or if the role is onsite and you only work remote. These are simple rules applied to yes/no questions. They've existed for fifteen years, they involve no artificial intelligence, and they're mostly fair: a company that can't sponsor your visa won't start because a human read your CV.

What about the AI features? I work in an ATS that ships new ones every quarter, so here's an honest inventory. Today, they summarize interviews, automate pipeline steps and rate CVs against the job description. That last one sounds like the famous robot, so it deserves a closer look. In practice, recruiters don't trust these ratings. They get fooled by basic keyword tricks, they bury interesting profiles and rank mediocre ones on top, so at best we treat them as a vague indication. They decide nothing. Any application remotely coherent with the role lands in my inbox, whatever its score.

So your CV, your written answers and your links end up in front of a human being, one with very little time per application, but a human who reads. Which brings us to the part of the process that actually eliminates candidates.

The real problem: applications written by AI

Remember the 95%. Most companies ask a few short questions when you apply: why this company, why this role, tell us about a relevant project. A recruiter reads every one of those answers, and 95 times out of 100, what I receive is the same paragraph, generated by the same tools, with the same vocabulary, the same sentence rhythm and the same polished emptiness. After a few hundred of these, you recognize AI writing in about a second.

Here's why that matters: the moment I identify a generated answer, I stop reading it. Partly out of fatigue, but mostly because that paragraph contains no information about you. It doesn't tell me how you think, how you express yourself or what you actually care about, so there's nothing left to evaluate, and I move on to the next application in the pile.

The same logic damages CVs. Candidates ask AI to optimize their CV for a filter that doesn't work the way they imagine, and the AI obliges with 70 skills, every buzzword in the industry and dense blocks of text. That CV may score well against a robot, but when it reaches me it's unreadable, and unreadable means discarded. The trap also catches honest candidates: you write a sincere answer, then run it through ChatGPT to clean up the syntax, and the tool sands away everything that sounded like you. What comes out is the same voice as everyone else's.

This is the real automated rejection happening today. Candidates hand their application to a robot in order to beat a robot that wasn't reading it, and eliminate themselves with the human who was.

Why volume doesn't work

There's a fair objection to everything above: if it takes hundreds of applications to get a single callback, automating the process is a rational response, and recruiters are poorly placed to complain about it. I understand that frustration, and parts of it are legitimate. Ghosting happens far too often. Screening questions are generic, including mine, though I keep them generic on purpose: they're quick to answer and they leave you free to say whatever feels most relevant about yourself. And no one evaluates a person fairly in a few minutes, something I've written about in Inside a recruiter's head.

The difficulty is that mass automation makes the situation worse for the very people using it. LinkedIn now receives over 11,000 applications per minute, a 45% increase in a single year driven largely by AI and auto-apply tools, and an average posting draws around 240 applications. Faced with that volume, every recruiter I know reacts the same way: less time per application, faster skips, and more weight given to the few signals that can't be faked. Two hundred automated applications land in two hundred piles where nearly everyone looks identical, add noise to a market that already has too much of it, and earn about a second of attention each. Volume has never been a good job-search strategy, and it has never worked as poorly as it does now.

What to do instead

• Apply to fewer roles, properly. Ten targeted applications with real answers will take you further than 200 automated ones. If you're sending hundreds of applications without a single callback, the volume is the problem, and organizing your search matters more than expanding it.
• Keep your CV standard and readable. One page, clear structure, the skills you actually have. The 20 seconds a recruiter gives it are the filter that matters, and everything in Anatomy of a good CV still applies, with no keyword stuffing required.
• Write the short answers yourself. Five to ten minutes, in your own words. Imperfect phrasing carries more information about you than a polished paragraph that says nothing, and bullet points or simple sentences are perfectly fine. Typos are the one thing to eliminate, so use a checker for grammar and keep your wording.
• Be specific rather than literary. For a company like mine, "I play competitive basketball and use your product's category daily + I want a small company at exactly your stage + you're building this function from scratch and I've done that before" beats any generated paragraph. Three lines, true, specific and yours.
• Use AI where it helps. Research, structuring your thinking, checking your grammar. I use it myself when I apply. The words a human will read should stay yours, and used that way, AI even becomes an asset worth mentioning in interviews.

Where this is heading

I won't pretend the current situation is pleasant for anyone. Candidates spend hours customizing applications that will get 20 seconds of attention, recruiters wade through machine-generated noise, and trust erodes on both sides. It's what happens when one side automates against an imaginary robot and the other side drowns in the result.

I don't expect it to last. Within a couple of years, AI intermediaries will probably be good enough to do real matching work: sparing candidates the grind of tailoring every application, sparing recruiters the noise, and connecting people to roles on substance rather than keywords. The current AI features in even the best ATS are nowhere near that yet.

In the meantime, one thing works entirely in your favor: the more automated applying becomes, the more a few lines written in your own voice stand out.

Enjoying this article?

Get the next ones straight to your inbox.

you@email.comSubscribe

FAQ

Does an ATS automatically reject my CV?

Only on binary criteria, through knockout questions: salary far outside the range, wrong location, a visa the company can't sponsor, an onsite/remote mismatch. These are simple rules with no AI involved. Everything else, your CV and your written answers, is read by a human, even if quickly.

Should I use ChatGPT to write my application answers?

Use it for research, structure and grammar, never for the words themselves. Recruiters recognize AI-generated text within a second and skip it, because it carries no information about you. A slightly rough answer in your own voice consistently outperforms a polished generated one.

Why do I never hear back after applying?

Volume, in most cases. An average posting now draws around 240 applications, many of them automated, so recruiter time per application keeps shrinking. Silence usually means buried rather than rejected by a machine. The fix is fewer, sharper, more authentic applications, targeted at roles that actually fit.