AI and the Residency Application¶
The Prompts This Guide Uses
Mock Residency Interview
Rehearses residency interviews one question at a time, then debriefs from your answers.
Have ready: the specialty and program type you are preparing for, the parts of your application to probe, and the question you most dread.
Personal Statement Critic
Critiques a personal statement you wrote, with questions instead of rewrites.
Works with any capable assistant in the Tool Directory.
This playbook covers the tasks of the application year where artificial intelligence (AI) genuinely helps, the one task where it mostly does not, and the rules that actually apply to you as an international medical graduate (IMG).
It deliberately stops short of application strategy. Which programs to apply to, how many, how to signal, what your statement should be about, and how your own record should be presented are questions for the Education Enhancement Department (EED) clinical advising team, who know you and know this school's outcomes. Application dates move every cycle, so this page links the Electronic Residency Application Service (ERAS) timeline rather than restating it (obstetrics and gynecology and emergency medicine programs take applications through ResidencyCAS instead, which publishes its own).
The short version
The Association of American Medical Colleges (AAMC) permits AI for brainstorming, proofreading, and editing. The Educational Commission for Foreign Medical Graduates (ECFMG) advises against letting it write your statement. The Electronic Residency Application Service (ERAS) does not run AI detection on applications, but in a published survey of program directors in one specialty, a majority regarded any chatbot use in a personal statement as unethical, so the private standard you are judged against may be stricter than the published rule. Your Education Enhancement Department (EED) clinical advisor owns strategy; this page covers the tools.
The Five-Minute Version
Where AI Genuinely Helps¶
Interview Practice, the Strongest Use on This List¶
This is where AI earns its place in the application year, and it matters most for students who do not have a network of physicians to practice with. An assistant can run a mock interview that asks real follow-up questions, stays in role, and debriefs you afterward on what was vague, what was rehearsed-sounding, and what a program would probably probe.
Ask for one question at a time, answer out loud before reading further, and have it push back rather than praise. Ask for the follow-up you would least like to be asked. Practicing the answer to "walk me through this gap in your record" ten times in private is worth more than any drafting the same tool could do for you. The Mock Residency Interview prompt sets this up: it stays in role, asks one question at a time, and debriefs you from your own answers.
Program Research and Keeping Track¶
Assistants are useful for organizing: building the comparison sheet, summarizing what a program publishes about its curriculum, drafting the tracker that keeps deadlines and interview invitations straight.
Check every program fact against the program's own page. Models state program details fluently and often get them wrong, including details that change every year, so AI is most useful for organizing what you gather rather than as the source of it.
Your Own Draft, Interrogated¶
The useful role on a personal statement is critic, not author. Once you have written a draft, ask an assistant what is missing, where you told the reader something instead of showing it, which paragraph is doing the least work, and what a program director would want to ask you after reading it. Then rewrite it yourself. The Personal Statement Critic asks those questions, quoting your draft, and never rewrites it for you.
That order matters. A draft you wrote and had criticized reads like you. A draft the model wrote and you edited reads like the model, and it puts you inside ECFMG's advice against relying on AI to write it.
The Other Pieces¶
Experience descriptions, thank-you notes, and scheduling correspondence carry far less of your voice and far less risk. Accuracy matters just as much in these, because every fact about you is one a program may ask you about. For an experience description, that means giving the assistant only your facts (your role, the setting, what you did, and one concrete result) and asking it to fit them to the character limit without adding any.
Patient Stories in Your Statement¶
Personal statements are often built around a patient encounter, and an assistant can critique how you tell that story. The version to work on is the de-identified one, which is also the one to submit: a patient story told without identifying details is the professional norm, and the story's force is in what you learned.
Do what you already do on rotations: abstract first. An age band, a presentation pattern, no names, no dates, no institution, nothing that could identify a person. The story you tell in a statement should already be abstracted to that level anyway, because it is going to be read by strangers.
What the Rules Actually Say¶
Three organizations matter here, and their positions are different from each other and from what circulates online.
| Body | Position on AI in your application |
|---|---|
| Association of American Medical Colleges (AAMC), which runs ERAS | Acceptable for brainstorming, proofreading, or editing; the final submission must be your own work |
| Educational Commission for Foreign Medical Graduates (ECFMG), which certifies you | Advises against relying on AI to write personal statements; advisory, not a sanctioned policy |
| National Resident Matching Program (NRMP) | No position on AI use; requires that everything you submit be complete and accurate |
- The AAMC permits AI with a limit. Its personal statement guidance states that "the use of AI tools is acceptable for brainstorming, proofreading, or editing the personal statement, but the final submission should represent your own work." The AAMC also investigates suspected plagiarism and may report substantiated findings to the programs you apply to, now and in later seasons. (AAMC personal statement guidance)
- ECFMG advises against AI-written statements. Its applicant guidance says plainly: "DON'T rely on artificial intelligence (AI) to write your personal statements," alongside the advice to show your statement to others for feedback and proofreading "but not for their rewrites." This is advisory guidance rather than a policy with its own sanction, and it applies to you specifically as an IMG. (ECFMG Personal Statement Do's and Don'ts)
- The NRMP has no position on applicants using AI, and says so directly: it has no role in the application or interview process. Its obligations are about accuracy. Information you provide must be complete and accurate, and a fabricated experience is a violation whoever or whatever wrote it. (NRMP Policy Corner: Utilization of Artificial Intelligence)
Obstetrics and gynecology and emergency medicine programs, including combined emergency medicine programs, take applications through a separate service, ResidencyCAS, instead of ERAS. If you apply through it, its own applicant attestation and personal statement requirements take the place of the AAMC's guidance; ECFMG's advice and the NRMP's accuracy rule apply either way.
Two corrections worth having, because both circulate widely:
Two things you may have read that are not true today
A stricter AAMC sentence, that a statement must not be "the product of artificial intelligence," is quoted in articles from 2024 and is not in the AAMC's current guidance. Do not plan around a rule that no longer says that.
And the ERAS pipeline does not run AI detection on your application. The AAMC states that its platform does not use AI to analyze, sort, or evaluate applications, and that the interview platform it works with does not attempt to determine whether essays were written with AI tools, specifically to avoid biases that fall hardest on some applicants. (AAMC on its own AI use)
The Part Worth Thinking about Before You Decide¶
Taken together, those two corrections give an honest picture that is uncomfortable but useful. Nobody is reliably detecting AI authorship: ERAS does not try, and studies consistently find that readers cannot tell AI-written statements from human ones. So the question is not whether you would be caught.
The question is what your reader believes. In a published survey that asked program directors directly, a majority regarded any use of a chatbot in writing a personal statement as unethical, while the AAMC's published rule permits brainstorming and editing. That survey is small and covers one specialty, so do not treat it as the view of every program, but the direction is clear enough to plan around: you may be judged against a stricter private standard than the published rule.
The practical consequence has nothing to do with rules. A statement you did not write is one you cannot defend in an interview, in a room with someone who has read several thousand of them, and generic is the most common failure mode of AI prose. The strongest reason to write it yourself is that it works better.
For Faculty Writing Letters¶
The AAMC addresses writers of a letter of recommendation (LOR) directly and permissively: generative AI "has the potential to support your letter-writing process, including crafting initial drafts of the LOR and editing its content and tone. Regardless of how you use genAI, you remain the author of the letter and are responsible for its content, accuracy, and the assessment it conveys."
Two things specific to letters. A letter drafted from a thin prompt reads generic, and a generic letter does less for the student than a short specific one. And a letter is strongest built on your own observations: ask for the structure, then add grades, evaluations, rank, and the specific moments yourself in the final version, since those are what only you can supply.
Related: Module 6: Clinical Contexts for turning a real case into a teaching abstraction, and Shelf and NBME Score Reports for the exams that precede this year.
Content last reviewed October 12, 2026. Review dates are maintained as described on the About page.