AI job applications: do they work, and do recruiters spot them? — jobbjobb
AI job applications: do they work, and do recruiters spot them?
Do AI-written job applications work, and do recruiters notice the AI? Here's what actually decides it, what works, and how to use AI without getting filtered out.
Arild Langtind··6 min read
When ChatGPT launched in late 2022, everyone said job applications were "done" as a genre. Suddenly anyone could press a button and get out perfect prose. Three years on, the picture is both more interesting and more nuanced: AI has changed how people write applications, but not what actually gets them through.
This post sums up what actually gets an AI-written application through, based on what recruiters flag as problematic, what research on AI in recruitment shows, and what works in practice.
The short answer
Yes, AI applications work. A large MIT Sloan study of nearly 481,000 job seekers found that AI assistance on the CV produced 7.8% more job offers and 8.4% higher pay. But the effect doesn't come from the AI writing better than you. It helps you write faster, structure better, and focus on what's relevant for this role. The difference between a good AI application and a bad one almost always comes down to how much you've put into the prompt and the editing.
And yes, recruiters often see that AI was involved. But that's usually not what they react to. What they react to is generic sentences without substance, clichés ("I'm an outgoing and structured person who loves a challenge"), and the fact that the application could just as well have been sent to any other role.
What recruiters actually react to
Experienced recruiters consistently flag the same problems, whether the application is AI-written or not. Four things come up again and again:
Generic opening. "I am hereby applying for the position of [title] at [company]" is the sentence most recruiters say "throws me out of the text immediately".
No connection to the role. Candidates listing their qualities without showing they've read the posting.
No concrete examples. "Experience with project management" without saying which projects, how many people, what the outcome was.
Length. Applications longer than one page rarely get read to the end.
None of these are AI-specific problems. They're writing problems. But AI makes it easier to produce a lot of generic text quickly, so they show up more often when people use AI without editing.
The typical AI fingerprint
If a recruiter actually is going to recognise AI style, they're typically looking for patterns like these:
Three-item lists of personal traits. Classic LLM output: "I am structured, solution-oriented, and collaborative".
Mirroring the posting's language without context. The posting asks for "data-driven decision-making", and the application contains that exact phrase three times without saying which data you've actually worked with.
"It would be a pleasure to contribute"-style phrasing. Overly polite closings nobody actually writes naturally.
Too-perfect grammar with subtly off word choice. AI English (and AI Norwegian) tends to use words that are technically correct but uncommon ("to initiate" instead of "to start", "endeavour" instead of "effort"). For a walkthrough of how to avoid this in practice, see the guide to ChatGPT cover letters without bot style.
None of these are disqualifying in themselves. But they signal "lightly edited", and that's what lowers your score — not the AI itself.
How to use AI without getting filtered out
Here's what we've seen work best, based on applications that have actually led to interviews and offers:
1. Feed the AI concrete material from your own life
The most common mistake is to write a prompt like "write a cover letter for a developer role". What you get back is the average of all developer applications in the training data — i.e., something no one wants to read.
What works: give the AI three things:
The full job posting (not just the title).
Your CV or a short version of your career history, with numbers, projects, responsibilities. Need to tighten your CV first? See how to write a CV that actually gets read.
2–3 concrete examples of things you've done that are relevant to this role.
The AI is never better than the raw material you give it. With rich material it writes an application that's yours, and just helps you put it together. Here's how to phrase the prompt itself →
2. Edit out anything that could have been written for anyone
After you've got the first draft, go through and ask of each sentence: "Could this sentence sit in an application to a completely different role?" If yes, rewrite it so it couldn't — or delete.
This is the simplest rule for filtering out AI-generic text. It's about matching against the posting, and we go deeper into that in the post on why matching is everything.
3. Keep your own voice in the opening and closing
Even if the AI writes the body, the first and last paragraph should be your own words. That's where the recruiter forms a first impression, and it's the easiest place to sound like a template if you let the AI run it.
A good opening is often a concrete observation or a specific reason: "I've been following the company since you launched X in 2024, and especially the way you ..." — not "It is with great interest that I apply ..."
4. Read it out loud before sending
If you stumble over a sentence reading it aloud, so will the recruiter. AI-written text is often grammatically correct but has a flat rhythm. Reading aloud reveals it instantly.
The real question isn't "do they notice"
It's "have you delivered something worth reading". If the answer is yes, it doesn't matter which tools you used along the way. If the answer is no, the fact that your AI is new and fancy doesn't save you.
AI has lowered the bar for producing an application. That means the average of the application piles recruiters go through has become more uniform, not better. There's room to stand out, and that room is about specific content and personal voice. Same as before.
How jobbjobb helps
We built jobbjobb because we felt general AI tools like ChatGPT were missing two things when it came to job applications: knowledge of your specific career, and the ability to analyse the role you're applying for.
When you upload your CV once and paste in a job posting, jobbjobb does what the job requires: links your actual experience to the requirements in the posting, suggests concrete examples you can highlight, and writes a draft that's yours — not an average text. You edit, we do the grunt work.
Do recruiters spot AI-written applications?
Experienced recruiters recognise typical AI patterns, but what they react to isn't that AI was used. It's that the application is generic, lacks concrete examples, and doesn't show you've understood the role. An AI application that is personal and specific doesn't stand out negatively.
Is it allowed to use AI in job applications?
Yes. There's no law in Norway banning the use of AI in cover letters or CVs. Some employers may ask you to complete tests or assignments without AI during the process, but the application itself is yours to write — with or without tools.
Should I tell them I used AI?
No, not as a point of its own. It's about as relevant as saying you used spell-check. What matters is that the content is true and that you stand behind it.
Can ATS systems reveal that an application is AI-generated?
ATS systems (Applicant Tracking Systems) parse your text for keywords and structure. As of 2026 they have no reliable way to classify text as AI-generated. Dedicated AI-detector tools exist, but they produce many false positives and are very rarely used in recruitment.
A practical guide to writing cover letters with ChatGPT that actually get answered. Concrete prompts, before/after examples, and the five mistakes most people make.
Heading into the interview for your first summer job? Here's how to prepare, how to dress, how to answer the common questions — and what the boss is NOT allowed to ask you about.
Concrete job interview prep for 2026: what you must know about the company, how to answer with the STAR method, what to do with the salary question, and what an employer is not legally allowed to ask.