Do AI résumés and cover letters give you an advantage?

Every application you send now competes with a pile of polished ones. Most of that polish came from the same handful of language models. If you are wondering whether an AI-written résumé or cover letter gives you an advantage, the short answer from the research is that it mostly gives you the same thing everyone else has.

That is not the same as saying it does nothing. The studies point in more than one direction, and the details matter more than the slogan.

Cheap writing stopped being a signal

Anaïs Galdin and Jesse Silbert studied applications on Freelancer.com before and after language models arrived. Before, employers paid more for workers whose applications were customized to the job. After, they did not. A customized cover letter used to cost effort, and the effort told you something. Now it costs a prompt.

In the authors' model of the market, that makes hiring less meritocratic: workers in the top fifth of the ability distribution are hired 19% less often, and workers in the bottom fifth 14% more often. The paper has the model, and Tuck's write-up is the readable version.

This is the argument we care about. When anyone can produce a good letter, the letter stops separating the people who can do the job from the people who cannot.

The best case for AI writing help is about clarity

Emma Wiles, Zanele Munyikwa, and John Horton ran a field experiment with nearly half a million jobseekers on an online labor market. Jobseekers who got algorithmic writing assistance on their résumés were hired 8% more often at 10% higher wages, and employers were no less satisfied. The résumés had fewer errors and were easier to read.

That result is real, and it cuts against "AI gives you nothing." Our reading is narrower. The gain came from clearer writing. Once every applicant has clear writing, clarity stops distinguishing anyone. That is our inference, not something the paper claims.

The one measured advantage is a screener bug

Jiannan Xu, Gujie Li, and Jane Yi Jiang found that language models used to screen résumés prefer résumés written by the same model. Self-preference ran from 68% to 92% across the models they tested. In their simulations, candidates who used the same model as the screener were 23% to 60% more likely to be shortlisted than equally qualified candidates who wrote their own.

If you have heard that AI résumés sometimes "beat the ATS", this is the documented version. It is an advantage, and it has nothing to do with being better at the job. The same paper reports that simple interventions cut the bias by more than half, which is what you would expect people who build screening software to do once they know about it. Anyone counting on this is renting a bug that has a published fix.

We looked for rigorous studies showing that AI-written résumés confuse applicant tracking systems, and did not find any. If you know of one, we would like to read it.

Recruiters are not fooled as much as you hope

An IEEE-USA piece on a survey of 600 US hiring managers reports that 19.6% would reject an applicant for using generative AI on a résumé or cover letter, and 33.5% say they can spot an AI-written résumé in 20 seconds or less. The same article quotes tech leaders who see no problem with engineers using AI to word their résumés, as long as the content is accurate.

So the reviewer is neither hostile nor naive. Generic text loses. Accurate text is fine. Neither of those makes AI text an advantage.

They also have a volume problem that would make anyone suspicious of polish. LinkedIn reportedly sees 11,000 applications a minute, and Greenhouse, an applicant tracking vendor, reports that applications per role roughly doubled between 2022 and 2025, from about 115 to 244, in its AI in hiring reporting. That is a vendor survey, so weigh it as one. Even so, a recruiter reading 244 applications for one role is going to notice which ones read like the other 243.

What predicts good hires

Sackett and colleagues reanalyzed decades of selection research in a 2022 paper. In their corrected estimates, structured interviews were the strongest single predictor of job performance, at a validity of .42. Cognitive ability tests came in at .31.

We are not claiming that paper says anything about cover letters. Our takeaway is that the strongest predictor on that list is a structured conversation, and a polished document is not one.

What Advant does

Advant does not write your application with a language model. You build your résumé once in Advant's résumé builder, with versions for the kinds of roles you want, and you write your own cover letter template. For each job, Advant picks the résumé version whose keywords match the posting, counting a match in the job title three times as much as a match in the description, and it shows you which version it chose and why. The cover letter gets the company and role filled in, and nothing else. Advant does not invent text.

A language model is used in two places: reading the résumé you upload into the builder, and suggesting skills for your profile. Neither of them writes to an employer.

That is a deliberate choice. The words an employer reads are yours, and so is anything they can check.

What to do instead

• Lead with something a reviewer can verify: a repo, a pull request, a system you shipped.
• Write one specific sentence about why this team, in your own words.
• Fix the errors. Clear writing helped in the Wiles study, and it is still worth doing.
• Apply to roles where you fit, and fewer of them.
• Prepare for the structured interview, since that is where the evidence says the signal is.

Limits of this evidence

Most of these studies come from freelance marketplaces or simulations, not software engineering hiring at a company. The Greenhouse figures come from a vendor survey, and the IEEE-USA piece reports on a survey rather than presenting it. We read abstracts and summaries of some of these papers rather than every table, so check the sources before you quote a number.

Sources

• Galdin and Silbert, Making Talk Cheap: Generative AI and Labor Market Signaling, and the Tuck School of Business write-up.
• Wiles, Munyikwa, and Horton, Algorithmic Writing Assistance on Jobseekers' Resumes Increases Hires, published in Management Science.
• Xu, Li, and Jiang, AI Self-preferencing in Algorithmic Hiring, published at AIES.
• Sackett, Zhang, Berry, and Lievens, Revisiting meta-analytic estimates of validity in personnel selection, Journal of Applied Psychology, 2022.
• IEEE-USA InSight, What Tech Hiring Managers Really Think of AI-Created Resumes and Cover Letters.
• CNBC, Recruiters are drinking through a fire hose of job applications, and eWeek, Job seekers flood LinkedIn with 11,000 applications a minute.
• RecTech Media on Greenhouse, More applications, fewer recruiters, and the Greenhouse 2025 AI in Hiring Report.

For how Advant files applications, see auto apply, and for a checklist on comparing tools, nine questions to ask before you pay.