Should you use AI to write your resume?

01

What does AI do well on a resume?

Harvard’s Mignone Center for Career Success describes the role in the same terms: generative AI can be a useful tool in the editing process, helping you brainstorm revisions to bullet points, work in relevant keywords from a job description, or otherwise improve what you already have. The last phrase carries the weight. There has to be something of yours to improve.

02

What does AI do badly?

It invents. The U.S. National Institute of Standards and Technology calls this confabulation: confidently stated but erroneous or false content. Georgia Tech’s career center describes how it shows up in application documents: AI tends to repeat wording from the sources it is given and in many cases will create statements that are not based on fact or representative of your experiences.

It inflates. “Helped” can become “led,” and “used” can become “expert in.” A team result becomes yours.

It flattens. Harvard’s guidance says generative AI should not be the primary author, not least because its output will likely be very generic. Carnegie Mellon’s career center strongly discourages using chatbots to generate the majority or entirety of your resume bullet points.

It cannot weigh your experience. It does not know which project mattered in your field, what your manager would say about you, or what you want to do next. Those choices shape a resume more than wording does.

It does not see your finished file. A chatbot returns text. Layout, fonts and the file an employer’s system will read are still yours to get right, following the rules for an ATS-friendly resume.

03

Why is invention the risk that matters most?

A chatbot asked to make a bullet “more impressive” completes the pattern of an impressive bullet, and that pattern often includes a number and a tool.

The invented line reads better, which is the problem. It gets caught later, in one of four places.

• The interview. “Tell me how you measured that 20%.”
• A skills check. A test or a practical question on the tool you supposedly used.
• Verification. Titles, dates and credentials can be compared with records, as described in what shows up on a background check.
• The job. You are hired for skills you do not have.

The consequences fall on you, not the tool. The U.S. government’s Declaration for Federal Employment states it plainly for federal jobs: a false statement on the declaration or the materials attached to it may be grounds for not hiring you, or for firing you after you begin work. Private employers set their own policies, and many application forms end with a similar statement that you sign.

04

What is a safe workflow for using AI on your resume?

1. Write the facts yourself. Build a master resume with every role, the dates, what you did and any result you can source. Harvard’s first step is the same: begin with your existing resume or as many details as you can pull together. If you are starting from nothing, how to write a resume covers each section.
2. Find your own results first. A chatbot cannot know your outcomes. The questions in accomplishments for a resume help you recall them.
3. Remove identifiers. Take out your name, contact details and anything confidential from an employer. Harvard’s caution is not to share personal or proprietary data with generative AI, and Georgia Tech’s is to remove personal identifying information from anything you provide.
4. Set the rules in the prompt. Only your facts, gaps marked instead of filled, no new numbers.
5. Work one section at a time. Harvard suggests entering your information section by section or bullet by bullet for precise adjustments. Small pieces are also easier to check.
6. Ask for a list of changes. You should see the original and the edit side by side.
7. Verify every line. Numbers, tools, titles and dates first. Reject any edit that makes you sound more senior than you were.
8. Finish in your own file. Paste the text into your resume, fix the layout, and read it aloud.

Act as an editor. Below is one section of my resume.
1. Tighten each bullet to under 25 words, starting with a verb.
2. Fix grammar, tense and inconsistent formatting.
3. Where a bullet describes a duty with no result, do not invent one. Ask me a question that would help me add it.
4. Do not add skills, tools, numbers, titles or certifications. Do not make my role sound more senior.
Return a table: original bullet, edited bullet, what changed, and your question for me if you have one.

Section:
[paste, with personal details removed]

Use Harvard’s test as your final one: be able to speak to every line of your resume if asked in an interview. For adapting the finished resume to a specific posting, the four-step sequence is in ChatGPT prompts to tailor your resume.

05

When should you not use AI on your resume?

• The application says not to. Harvard advises reading all application instructions, because some organizations have rules about whether AI can be used in preparing materials.
• You have no draft. With nothing to work from, a chatbot writes a resume for an imaginary person. Do the thinking first.
• Precision is the point. Licenses, clearances, publications and clinical or legal credentials have exact names and numbers. Type those yourself.
• Writing is the job. If the role is judged on your writing, the resume is a sample of it.

Georgia Tech’s career center puts the principle in one sentence: the documents you create and submit should be your own.

FAQ

Frequently asked questions