Should I use AI to write my resume?

On this page

Short version
Who it's for
What to do
Your options
Beliefs checked
Questions to ask

The short version

Yes for the wording, no for the facts. The evidence leans toward writing your own content, using an AI tool such as ChatGPT to fix and tighten it, then editing the result so it sounds like you.

In a 2023 field experiment with nearly half a million jobseekers on an online labor market (Wiles, Munyikwa and Horton, NBER), jobseekers given algorithmic writing assistance on their resumes were 8% more likely to be hired in their first month on the platform, with no evidence that employers were less satisfied. The 2023 Wiles, Munyikwa and Horton experiment ran on an online labor market, among people that a 2024 dissertation reporting what appears to be the same experiment calls new labor market entrants (people just entering the job market). So treat the size of its gain as uncertain for you, especially if you have a long career.

Of the 4 options below, using AI to polish a draft you wrote has the best support we hold. Letting AI draft it has weaker support. Writing with no AI may cost you some hires. We hold no evidence that hiding text helps. Gap: we hold no comparable test with US employers hiring for regular jobs, and the parts of the 2023 paper we hold don't say how its writing assistance compares with today's AI chatbots.

On this page, a "vendor survey" is one run by a company that sells hiring or resume products. It reports what people told that company, not a measured result.

Who this is for

You're writing or updating a resume (CV) or a cover letter, and you're wondering whether AI tools such as ChatGPT will help you or hurt you. You may also have been told to use AI to add keywords from job descriptions to your resume. 77% of 1,200 US job seekers said they had used AI in their search in the past year (Greenhouse, 2026 vendor survey). In that 2026 Greenhouse survey, 60% had changed or considered changing how they write so their application isn't mistaken for AI.

What to do this week (our advice)

Write down your real experience first, then let AI help with the wording, so the content stays yours. If you already have a resume, start from it.
Run an error check with an AI tool before you send anything. More errors were linked to fewer hires in the 2023 Wiles, Munyikwa and Horton paper (option 2).
Edit every AI draft before you send it, and write the personal parts of a cover letter yourself. More editing time was linked to more offers in the 2025 Cui, Dias and Ye study (option 3). AI hardly improved personal motivation in the AI cover-letter study reported by the GLOBE academic newsletter. Hiring managers in the survey reported by HRreview name impersonal language as a sign of AI (see the second popular belief below).
Check every line is true before you send it. Employers in iprospectcheck's 2026 survey report withdrawing offers over resume fraud or credentials they couldn't verify (see the questions below).
Use the posting's own words naturally, where they truly apply, without stuffing in keywords. 76% of the 25 US recruiters vendor Enhancv interviewed mentioned natural use of keywords as something they value most.
Make sure your past work record holds up: on one platform, employers leaned more on past work experience after an AI cover-letter tool launched. On one online platform, employers shifted toward signals such as past work experience after an AI cover-letter tool launched (Cui, Dias and Ye, 2025).
Don't hide text for screening software. In a 2025 Greenhouse vendor survey, hiring managers reported catching it (option 4).

Your options

1. Write it yourself, without AI

What happens. You draft and polish every word on your own.

What it gives. Full control over every word.

What it gives up. Possibly some hires, at least on an online labor market like the one studied. Jobseekers given writing assistance were hired more often than those without it (the 2023 Wiles, Munyikwa and Horton experiment on an online labor market). A 2024 dissertation reporting what appears to be the same experiment describes the people studied as new labor market entrants.

What the evidence says. Recruiters who knew a cover letter was written without AI rated high-quality, fully human-written letters more positively (an experiment reported by the GLOBE academic newsletter, page read October 2026). Gap: we hold no study comparing fully human resumes with AI-assisted ones at US employers.

2. Use AI to check and polish a draft you wrote

What happens. You write the content, then ask an AI tool to fix errors and tighten the wording.

What it gives. A chance to catch errors before an employer does. Each additional percentage point of errors per word in a resume (by our arithmetic, one more error per 100 words) was linked to a 3% lower probability of being hired (the 2023 Wiles, Munyikwa and Horton paper, from observational data on a large online labor market; the paper's arXiv version lists the first author as van Inwegen).

What it gives up. Gap: we hold no study of any cost of using AI only to proofread. ChatGPT sometimes made texts more complex and less readable, says the AI cover-letter study reported by the GLOBE academic newsletter. So read the result back.

What the evidence says. Jobseekers who received the writing assistance also had 7.8% more job offers over the experimental period than the control group, who did not get it (the 2023 Wiles, Munyikwa and Horton experiment). A 2024 dissertation in MIT's online repository reports what appears to be the same experiment (same size, same 8% result). That 2024 dissertation says the resumes given the help had fewer errors, were easier to read, and were AI-improved but not entirely written.

3. Let AI such as ChatGPT draft it, then edit it into your own words

What happens. You give an AI tool your work history and the job posting, let it write a draft, then rewrite it.

What it gives. A head start on the wording. Gap: we hold no test of resumes drafted whole by AI; what follows in this option is about cover letters. Access to an AI cover-letter tool raised the chance of a callback (an employer getting back to you) by 0.43 percentage points, about 6% above a 7.02% baseline (a 2025 study by Cui, Dias and Ye on a large online labor platform). In that 2025 Cui, Dias and Ye study, actually using the tool raised the chance by 3.56 percentage points, a 51% rise. Both 2025 Cui, Dias and Ye results were significant only at the 10% level, the looser test of whether a result could be chance.

What it gives up. The edge may fade. The callback boost appeared in the first two months after the tool launched, then tapered off (the 2025 Cui, Dias and Ye study, significant only at the looser 10% level).

What the evidence says. Editing seems to matter. In the 2025 Cui, Dias and Ye study, most AI-generated cover letters were sent with little or no change. Among the same workers, noticeably more editing time (one standard deviation, a measure of spread) was linked to a 0.31 percentage point higher chance of an offer (the 2025 Cui, Dias and Ye study). Offers were rare in the 2025 Cui, Dias and Ye study, about 0.6% of AI-assisted applications by our arithmetic.

Generative AI helps job seekers write better cover letters but does not increase their chances of getting a job, concludes research by four Tilburg University economists, as reported by IO+ (page read October 2026). Better letters did not bring more interview invitations, says the GLOBE academic newsletter's report on an AI cover-letter study (page read October 2026). The GLOBE academic newsletter's report also says personal motivation and clarity of writing, the things recruiters care about most, hardly improved. That report may be about the same research as the Tilburg study, so don't count them as 2 findings.

These cover-letter sources point different ways. The 2025 Cui, Dias and Ye study found an early, weakly significant rise in callbacks, while the study the GLOBE academic newsletter reported found no rise in interview invitations. We can't settle that difference, so don't count on a cover-letter gain.

4. Hide keywords or instructions for screening software

What happens. You add text a reader can't see, meant to get your resume past AI filters (a "prompt injection").

What it gives. Gap: we hold no study showing that hidden text improves an applicant's chances.

What it gives up. In a 2025 Greenhouse vendor survey of 1,236 recruiters and hiring managers in the US, UK, Ireland and Germany, 65% of hiring managers said they had caught some deceptive AI use. In that 2025 Greenhouse survey, 22% said they had caught applicants hiding prompt injections in resumes, as part of that 65%. Gap: we hold nothing on what happens to an applicant who is caught.

What the evidence says. Some applicants try it anyway, though Greenhouse's 2 vendor surveys don't settle how many. 41% of 1,200 US job seekers admitted using prompt injections, hidden text designed to bypass AI filters (Greenhouse, 2025 vendor survey). 62% of 1,200 US job seekers tried tactics to get past automated screening (Greenhouse, 2026 vendor survey). Greenhouse says a third of those added hidden text meant for an AI reviewer, though the wording of its 2026 survey report leaves unclear which group "those" means.

Two popular beliefs, checked

"Using AI on your resume gets you rejected"

Not as a rule, but AI help and an AI-written result are judged differently. 52% of 600 hiring managers said using AI for proofreading or drafting support is acceptable (TopResume, 2025 vendor survey). In that 2025 TopResume survey, 19.6% said they would reject a candidate with an AI-generated resume or cover letter. TopResume, reporting that 2025 vendor survey, says the expectation is that the final product must sound human, show real effort and reflect the individual behind the words.

Jobseekers given writing assistance were hired more often, and the authors found no evidence that employers were less satisfied (the 2023 Wiles, Munyikwa and Horton experiment on an online labor market).

In iprospectcheck's 2026 survey of 1,500 business managers and company owners, AI-generated resumes were the most common type of fraud employers reported running into, at 14.0%. Employers in that 2026 iprospectcheck survey count AI-generated resumes among the fraud they run into, so a resume that reads as AI-written may be treated with suspicion. The parts of the iprospectcheck survey we hold don't say 14.0% of what, or separate AI-written from AI-edited resumes.

"Recruiters can always tell when AI wrote it"

Not shown, and the evidence is split. 33.5% of hiring managers said they can spot an AI-generated resume in under twenty seconds (TopResume, 2025 vendor survey). 73% of hiring managers said they can identify when a candidate has used AI on CVs or cover letters, citing impersonal language and overly polished formatting (a survey reported by the UK site HRreview, page read October 2026). In that survey reported by HRreview, 35% of those hiring managers said it hurts their view of the applicant. Recruiters were no better than chance at telling whether a cover letter had been written with AI (the experiment reported by the GLOBE academic newsletter).

These answer different questions. The surveys asked hiring managers about their own ability, while the experiment tested recruiters, and on cover letters rather than resumes. So don't rely on AI use going unnoticed, or on it being spotted. Fix what the hiring managers in the survey reported by HRreview say they notice: impersonal language and overly polished formatting.

Questions to ask yourself

Is every line true, and specific to me? Employers can act on credentials they couldn't verify, not only on resume fraud. 40.3% of 1,500 business managers and company owners said they had withdrawn a job offer because of resume fraud or credentials they couldn't verify (iprospectcheck, 2026 survey). In that 2026 iprospectcheck survey, 30.2% had found out about resume fraud only after the person was on the payroll. Gap: we hold no study of what happens when an AI tool adds details that are not true, such as skills you don't have, and nothing measuring how often background checks catch AI-written content as such.
Should I say I used AI? The only evidence we hold is about cover letters, in the experiment reported by the GLOBE academic newsletter. That GLOBE-reported experiment concerns letters recruiters knew were written without AI, not applicants who said they used it. Gap: we hold no test of telling employers you used AI on a resume.
Does it use the job posting's own terms? 76% of 25 US recruiters mentioned natural use of keywords, not keyword stuffing, as something they value most (vendor Enhancv's interviews, page read October 2026). 88% of employers agree that qualified high-skills candidates are vetted out (screened out) because they do not match the exact criteria in the job description (the 2021 Harvard-Accenture report on hidden workers, as quoted in The Register). In a 2024 Lake Forest College news report on a senior thesis, the student, Stoia, says that across over 500 applications, those with an existing resume tailored by ChatGPT saw a 14% increase in callbacks and those tailored by Google Bard a bit over 15% higher probability of an interview. Gap: we hold only the college's report, not the thesis, so we can't say how the applications were split or how reliable those differences are.
Will software (an applicant tracking system, or ATS) read it before a person does? Possibly, and you can't tell from these figures what a given employer does. So write for both a person and software. 23 of 25 US recruiters (92%) said their systems do not auto-reject (turn down automatically) resumes for formatting, content or design (vendor Enhancv's interviews, page read October 2026). In those Enhancv interviews, 8% had AI or content auto-rejection configured. All 25 US recruiters in vendor Enhancv's interviews said they use knockout questions when present: yes or no eligibility checks such as work authorization, required license and location. So answer those accurately.

57% of 1,399 full-time US workers with knowledge of their company's hiring practices said their company used AI in hiring (Resume.org, 2025 vendor survey, reported by CPA Practice Advisor). In that 2025 Resume.org survey, three in four of those companies allowed AI to reject candidates without human review: 35% at any stage and 39% only at initial resume screening.

AI models used as screeners consistently preferred resumes they had generated themselves over those written by people or by other models, even with content quality controlled (a controlled resume experiment reported in the paper "AI Self-preferencing in Algorithmic Hiring"). That experiment tested the models, not any employer's hiring, so we would not draft with AI on this alone.
Is it the resume or the cover letter? The strongest evidence for AI help is about resumes. For cover letters, the Tilburg University research reported by IO+ found no gain in the chance of getting a job, and the 2025 Cui, Dias and Ye study found the early callback boost faded.
What has AI changed for everyone, whichever option I pick? These studies suggest changes that could reach you whether or not you use AI. After the tool launched, the link between how tailored a cover letter was and getting a callback fell by 51%, and the link with getting an offer fell by 79% (the 2025 Cui, Dias and Ye study of one online labor platform). The 2025 Cui, Dias and Ye paper says that after the tool launched, employers shifted toward other signals, such as workers' past work experience. Employers had a high willingness to pay for workers with more customized applications before AI writing tools (large language models) were introduced, but not after (a separate 2025 paper, "Making Talk Cheap: Generative AI and Labor Market Signaling").
Am I a strong writer, and how far along in my career? AI helps less qualified applicants most, making candidates more evenly matched and harder for recruiters to tell apart, says Till Wicker, a researcher cited by IO+ in its report on the Tilburg University cover-letter research. Workers with higher pre-AI writing ability (how well they wrote before the tool) tended to spend more time editing AI drafts (the 2025 Cui, Dias and Ye study). Gap: we hold no measurement of AI resume help for career changers, workers over 40 or people returning after a long break.

How we checked this: this page is the output of a rigorous research pipeline run with AI tools: every factual sentence rests on a finding checked against the source's own words, and a person approved the page before it went up. How we check our facts. Updated November 12, 2026.