AI Cover Letters Lost 51% of Their Signal. Write the 'Why' Yourself

More than three-quarters of US job seekers now use AI to help them find work, so the cover letter you just generated reads a lot like everyone else's. Forbes reported on Oct 26 that 60% have changed or considered changing how they write applications to avoid sounding AI-generated, and that effort goes to the wrong problem. You don't have to stop using AI. You need to know which paragraph it can't write for you.

Do AI cover letters work?

AI cover letters work as a baseline, not as a way to stand out. In a Yale study of more than 5 million cover letters, having an AI writer raised callbacks at first. Then the link between a well-matched letter and a callback fell 51%, because everyone had the same tool.

The study comes from Jingyi Cui, Gabriel Dias and Justin Ye at Yale. They looked at over 5 million cover letters sent to 100,000 jobs on Freelancer.com. On April 19, 2023, Freelancer launched the "AI Bid Writer," a tool that automatically writes a cover letter tailored to the job description. That launch gave the researchers a before-and-after to compare.

The key term is textual alignment: how closely a cover letter's wording matches the job posting. Here's what they found:

• Getting the tool raised alignment and raised callback likelihood.
• After launch, the correlation between alignment and callbacks fell by 51%.
• Time spent editing the AI draft was positively correlated with getting hired.
• Employers shifted to other signals, including candidates' prior work history.

So AI didn't fail. It worked so well that it cancelled itself out.

Why did tailoring stop working?

Tailoring stopped working because a signal only counts when it's hard to fake. Once a tool can write a perfectly matched letter for anyone, a matched letter tells the employer nothing about who's good.

A signal is anything you send that's cheaper for a strong candidate to produce than for a weak one. Before generators, a letter that echoed the posting showed you'd read it closely and could write. Now it shows you pressed a button. Employers responded by looking elsewhere.

A second Freelancer study, "Making Talk Cheap" by Anaïs Galdin (Tuck) and Jesse Silbert (Princeton), estimates the cost. After AI, platform wages are about 5% lower across the board, hiring is 1.5% lower, and employers more often hire weaker candidates. Galdin describes the flood plainly: "Employers are receiving around 60 applications within a few minutes."

You feel that flood from the other side. Forbes reports that 81% of US job seekers have been rejected from at least one job without ever speaking to anyone at the company. And 67% of HR leaders say AI applications slow hiring down. You send a polished, tailored letter, it lands in a pile of polished, tailored letters, and the reply never comes.

The practical takeaway: matching the posting is now the minimum. It keeps you from being screened out. It won't get you picked.

What do recruiters actually weigh in a cover letter?

Recruiters weigh personal motivation and clarity most, and those are the two things AI barely improved. That's the main finding from Till Wicker's experiments at Tilburg University.

Recruiters from Philips, PwC, Rabobank and VodafoneZiggo scored cover letters written with and without ChatGPT. The AI versions looked more professional on the surface. On the dimensions recruiters care about, they hardly moved.

Two more Tilburg results are worth knowing:

• Recruiters were no better than chance at telling whether a letter had been written with AI.
• The researchers estimate growing AI use could reduce the quality of employer-employee matches by about one percent.

Forbes adds a caveat: the study used Dutch employers and an earlier version of ChatGPT, so don't treat it as a universal measure of today's tools. The direction still matches the Freelancer data, though. AI polishes the generic parts of the letter, and the generic parts aren't where hiring decisions get made.

Rewrite the middle, not the opening

Spend your editing time on the motivation paragraph, not the first line. In the Tilburg experiments, AI's gains landed on the intro and closing, which recruiters weigh least. Motivation, which they weigh most, barely moved.

Most people do the reverse. They agonize over "I'm writing to apply for..." and leave the middle paragraph exactly as the model wrote it: something about being "excited by the mission" and how their "skills align." That's the sentence a generator writes for everyone.

The cover letter motivation paragraph is the part that explains why you want this specific role at this specific company, and why now. It's also the one part of the letter only you have the material for.

How to write the motivation paragraph

1. Name one specific thing about this team. Pick a product decision, a public post from the hiring manager, or a line from the posting that isn't boilerplate. One thing, not three.
2. Connect it to something you actually did. A project, a problem you fixed, a system you built or migrated off. Name it.
3. Say what you want to do next. Explain why this role is the logical next step from that work, not just a nice job.
4. Cut any adjective that fits any company. "Innovative," "dynamic," "passionate" all go.
5. Run the competitor test. If the paragraph would still make sense sent to the company's biggest rival, rewrite it.

Here's an illustration. The generic version: "I'm drawn to your mission and believe my analytical skills align well with this role." The rewritten version: "Your posting says the team is moving reporting off spreadsheets and into a shared data model. I led that same migration for the finance team at my last company, and the part I liked most was teaching analysts to own their own models. I want to do that at a company where data is the product, not a back-office function."

The second version is checkable, specific and tied to real history. No generator writes it for you, because the facts only exist in your head.

This is how I'd split the work. Let Refolk draft the letter from your own work history and the posting, so the intro, closing and structure are done. Then put all your time into those four or five sentences of "why."

Stop trying to sound human. Make your claims checkable.

Stop trying to avoid sounding like AI, because recruiters can't tell anyway. Spend that effort making every claim easy to verify, since that's where employers have moved.

Forbes reports that 60% of job seekers have changed or considered changing how they write applications to avoid sounding AI-generated. Given that the Tilburg recruiters did no better than a coin flip at spotting AI letters, that's wasted work.

The riskier habit is padding. Greenhouse's 2026 Future of Hiring Candidate Report, which surveyed more than 2,800 job seekers in the US, UK, Ireland, Germany and Australia, found:

• 62% have tried tactics to get past automated screening.
• 54% used AI to figure out what the system is looking for.
• 40% added skills or keywords they don't have.

Now set that next to the employer side. Some 65% of hiring managers say AI makes verifying skills harder, and the Yale team found employers responded by leaning on prior work history. So a skill you made up gets tested at exactly the point employers now check most closely.

The motivation paragraph is your best defense here. When it points to a real project with real details, it backs up your resume instead of competing with it.

Most job seekers edit their letters alone

Most candidates have no one reviewing their cover letter, so they need a self-edit routine. In Refolk's index of professional profiles, US recruiters outnumber resume writers and career coaches by about 17.5 to 1.

These are loose title matches, so read them as scale, not exact headcounts. The shape is still clear. Almost 97,300 people in the US read applications for a living, and only about 5,550 help candidates write them. Greenhouse reports that 27% of job seekers have paid for a resume service, which leaves most people editing on their own.

The Netherlands row matters too. The Tilburg panel came from a recruiter market roughly one-eighteenth the size of the US one. That's one more reason to take Forbes' caveat seriously and treat the study as a direction, not a precise forecast.

How to edit an AI cover letter: a self-edit checklist

To edit an AI cover letter well, leave the structure alone, rewrite the motivation paragraph by hand, and make every claim match your resume. Work through this list in order:

• Check fit first. If you don't match the role, no edit will save the letter. Refolk scores how well you actually fit each posting, which tells you where your editing time is worth spending and where it isn't.
• Leave the opening and closing mostly as they are. Fix anything wrong, but don't polish. Recruiters weigh these least.
• Rewrite the motivation paragraph from scratch using the five steps above.
• Simplify. Tilburg found ChatGPT sometimes makes text more complex and less readable. Break up long sentences and swap formal phrasing for plain words.
• Match every claim to your resume. If the letter says you led something, the resume should show where and when.
• Delete any skill you can't demonstrate. Your work history is exactly what employers now check.
• Read it aloud once. Clarity is one of the two things recruiters care about most, and reading aloud is the cheapest test for it.

Are cover letters still worth it in 2026?

Yes, but only the edited version. A raw AI letter now proves almost nothing, while time spent editing is the one behavior the Yale study tied to getting hired.

To be fair, that's a correlation, not proof. Stronger candidates may just edit more because they care more or have more to say. But the practical advice is the same either way. The letter that gets read is the one with a specific, true reason for wanting the job, and you can only write that reason yourself.

If the application treats the letter as optional and your fit is weak, skip it and put the time into a role where you match. If the letter is optional and your fit is strong, write it. Five honest sentences about why this team, backed by your real work, is the strongest signal left in the document.

FAQ

Should I use ChatGPT for my cover letter?

Yes, for the structure, the greeting, the intro and the closing. Those are the generic sections AI improved in the Tilburg experiments, and they're the parts recruiters weigh least. Don't let it write the motivation paragraph. That's the part recruiters care about most, and it's the part AI barely improved.

Can recruiters tell if my cover letter was written with AI?

Not reliably. In the Tilburg study, recruiters from Philips, PwC, Rabobank and VodafoneZiggo were no better than chance at spotting AI-written letters. The real risk isn't detection. It's sending a generic letter that looks just like the other applications, or including a claim your work history can't back up.

How much time should I spend editing an AI cover letter?

The Yale study found that more time spent editing AI drafts was correlated with hiring success, but it doesn't give a magic number. Put almost all of that time into the "why this role" paragraph and into checking that every claim matches your resume. Leave the opening and closing mostly alone.

Does the 51% finding apply outside freelance platforms?

The data comes from Freelancer.com, where application volume is high: Galdin describes employers getting around 60 applications within a few minutes. But the mechanism holds anywhere tailoring is free: once every candidate can match the posting, matching stops telling good candidates apart from weak ones. The Tilburg results, from corporate recruiters, point the same way, with the caveat that they used Dutch employers and an older ChatGPT.