How to Research a College for a Why Us Essay
Where real school details live, how to use AI for research without inventing professors or courses, and a two-page check before any fact goes in.
How to Research a College for a Why Us Essay
Take every school fact in a why us essay from a page the college itself runs, and confirm it on two of its pages, one of them current for this term, before it goes in. AI chatbots can give you leads; Cornell and Northwestern list that use as acceptable. They are not sources.
Naming a specific course or professor only helps if the course runs this year and the professor teaches first-years. That check matters more if you use a chatbot for research, because a chatbot can produce a course title that sounds right and does not exist. This page gives you a source ladder, a verification routine for AI leads, and a constructed research log. When your facts are checked and drafted, the why-us supplement rubric reads how you connect them to yourself.
Every admissions page quoted below was checked on October 30, 2026. Policies change, so read each college's current page before you rely on it.
What colleges say about research, visits and AI
Six admissions offices, in their own words. Each row was checked on October 30, 2026.
Duke's distinction between demonstrated knowledge and demonstrated interest is set against Common Data Set ratings in do colleges care about why us essays.
| College | What it says about research | What it says about AI |
|---|---|---|
| Duke | "Explore the college websites. There is so much information there, including a lot geared towards current students that will give you insight into the college." | Not addressed on this page |
| Cornell | Lists as acceptable: "Using AI tools to research specific colleges or help you build your college list." | Lists as unacceptable: "Having AI tools outline, draft, or write your application responses, translate your responses from another language, or create works for a required portfolio." |
| Northwestern (updated 9/23/25) | Lists as constructive: "Researching colleges and admission terms" | Lists as problematic: using AI to write essays from prompts, to translate an essay, to complete an outline, or to rework a supplement written for another school |
| Wisconsin | Not addressed on this page | "We will not disqualify an applicant found to have used or suspected of using AI in their admissions essays, nor are we running essays through any system to detect if AI was used. However, we strongly discourage students from simply feeding AI a prompt for their essay, as it denies you the opportunity to help us better understand why you would be a good fit for UW–Madison." |
| Notre Dame | Not addressed on this page | "While an applicant may use artificial intelligence to assist with spelling and grammar review, or brainstorming support before writing, the content of all essays, short answer questions, activities, and any other material submitted must be an applicant's own work." |
| Tufts | Encourages "exploring the admissions website, registering for admissions events, reading Jumbo Magazine or our student blogs, and following us on Instagram" | "Tufts Admissions strongly discourages the use of generative artificial intelligence (AI) for writing college admissions essays and short answer responses." It adds: "The Admissions Committee expects to read an applicant's original ideas presented in their own words." |
Two of these rules need their full text. Northwestern's problematic list, word for word, is: "Using generative AI to write your personal statement or supplemental essays by inputting prompts and submitting the output as your own writing"; "Translating an essay written in another language"; "Providing generative AI with an essay outline and submitting AI's completed piece of writing"; and "Using generative AI to rework a supplemental essay that was written for one school for another school." Notre Dame's policy continues: "Submitting content generated, drafted, or substantially written by artificial intelligence tools or other text-generating software may result in the revocation of your offer of admission or potential expulsion." It also adopts the Common App Fraud Policy, which prohibits "intentionally misrepresenting as one's own original work … the substantive content or output of an artificial intelligence platform, technology, or algorithm."
Read these as information. Where a college publishes no rule, the certification inside the application is what you sign, so read it there. For other colleges, GradPilot's university AI policy directory collects published policies.
What the table settles for research: two of the six colleges name AI research of colleges as acceptable or constructive, and none of the six forbids it on these pages. The five that address drafting all ask for the writing to come from you, at different strengths: Wisconsin discourages feeding AI a prompt without disqualifying anyone, while Notre Dame warns of revocation. So research is the step where AI help is most widely accepted, and also where its mistakes are easiest to carry into a draft.
Where specifics live: a source ladder
Not every college page proves the same thing. This ladder is our guide to what each kind of page can and cannot show.
| Source | What it proves | How it goes stale |
|---|---|---|
| This term's class schedule | A course runs now, and who teaches it | Changes every term |
| Department undergraduate page | Requirements, first-year courses, advising, research for undergraduates | Update dates vary; check the page's date |
| Course catalog or bulletin | A course is part of the curriculum | Lists courses that may not run in your first year |
| Student organization directory | A club is registered | Inactive clubs can stay listed; look for this year's activity |
| Student newspaper archive | What students actually do and argue about | Each article is dated; use recent ones |
| Admissions blogs and virtual sessions | What the office chooses to emphasize | Posts can be years old and written for older prompts |
Our recommendation: a fact earns a place in your essay when it appears on two of these, and at least one is a dated or current page: the term schedule, this year's organization listing, a dated department page or a recent newspaper article. A course that appears only in the catalog is a lead, not a fact.
Duke's resources page makes the point about student-facing pages: much of a college's site is "geared towards current students." That is where the specifics live, because current students need them to plan a term.
Big public university vs a college built around one curriculum
The research route changes with the size of the institution and the shape of the prompt, but the proof standard does not.
At a large university, research the unit the prompt names. Michigan's Essay 2 asks about "the specific undergraduate college or school … to which you are applying at the University of Michigan" and "How would that curriculum support your interests?" A paragraph about football Saturdays or the size of the library answers a question Michigan did not ask. Start from that college's own pages: its first-year requirements, its advising pages, its student organizations. (Michigan's page states no cycle year and returns HTTP 403 to scripts; we read it through a text proxy.)
At a college organized around one curriculum, research the structure. Brown's academic fields page says students "sample courses across many disciplines before immersing yourself in one of these focused areas," and adds: "There are no quotas among concentrations and we do not admit students by academic interest." There the specific thing to research is how the curriculum lets you explore, not only which department you name.
For the school-by-school prompts, our Michigan and Brown supplement guides quote each one in full.
Using AI for research without importing its mistakes
The U.S. National Institute of Standards and Technology describes the problem in its Generative AI Profile (NIST AI 600-1, July 2024). It calls it confabulation: "a phenomenon in which GAI systems generate and confidently present erroneous or false content in response to prompts." It adds: "Risks from confabulations may arise when users believe false content – often due to the confident nature of the response – leading users to act upon or promote the false information."
In a why us essay, that risk has a specific shape: a professor who left, a course that was renamed, a lab that closed, a club that never existed. Each would survive the two-way swap test, because it is specific. Specific and false is worse than generic.
Our recommendation, as a routine:
- Ask for leads, not facts. Ask what kinds of programs, courses, labs or clubs to look for at that college. Treat every name in the answer as unconfirmed.
- Find each lead on two college-controlled pages. One should be current: this term's schedule, this year's organization listing, a dated department page.
- Log the URL and the date. The log below shows the format.
- Discard what you cannot find twice. Do not swap in something close. If the course is gone, the sentence goes.
- Write from the college's page and your own notes, never from the chatbot's wording. At Cornell, having AI outline or draft your responses is listed as unacceptable; at Notre Dame, the content "must be an applicant's own work."
- Check people last and hardest. A faculty name needs the department directory and evidence that the person teaches undergraduates. The sibling guide on naming professors, classes and clubs covers when a name helps at all.
The countercase: if a college publishes a list of first-year programs on one current, official page, such as a first-year seminar list for this year, one page can be enough. The two-page rule is for anything a chatbot, a forum or an older post told you.
A six-row research log
Use five columns: the fact, the URL, the date you checked it, what it lets you do, and which of your experiences it connects to. The last two columns are the start of your draft: a fact with nothing in column five does not belong in the essay.
Built your list on aid? The why us essay when you need financial aid shows a 30-minute version of this research.
Constructed example: a six-row log with two failed AI leads
Constructed example, written by us, not a real applicant's log. Fenwick College and everything in it are fictional, so the page names are descriptions, not URLs.
| Fact | Where confirmed (checked) | Lets me | Connects to |
|---|---|---|---|
| First-year seminar "Maps and Power" | Geography department page; this fall's schedule | Map who gets bus service | My route-timing project |
| Bike co-op run by students | Organization directory (active this year); student paper article from last spring | Fix bikes for other students | Three years at my town's repair shop |
| "Semester in the City" program | Admissions page; program page with this year's dates | Work in a city office in junior year | Council meeting I spoke at |
| Geography honors thesis | Catalog only | Unclear | Unclear. Held, not used: one page, no date |
| AI lead: "Prof. R. Ilunga teaches Coastal Geomorphology to first-years" | Not in the department directory; no such course on this fall's schedule | n/a | Discarded |
| AI lead: "Undergraduate GIS lab open to all students" | No college page found; one news story from a past year mentions a lab for graduate students | n/a | Discarded |
Three facts survive, and the essay can afford two. Notice the honors-thesis row: it is real enough to be in the catalog, but not confirmed for this year, so it stays out. The two AI leads were not obviously wrong. They were plausible, and that is exactly why the check exists.
If you never visited
You do not need a visit to write this essay. Duke's resources page says: "Don't worry about visiting the campus if you can't afford it or if it's too far away!" It asks for "demonstrated knowledge" instead of "the number of times they have visited, met, or contacted us." Tufts's current Short Answer Prompt #1 asks how you "learned about and engaged with Tufts during your college search process," and the same page names the admissions website, events, Jumbo Magazine and student blogs as ways to learn.
Write from reading, and say so plainly if it helps. Constructed example, written by us, not a real applicant statement: "I have not visited Fenwick. I have read this fall's geography schedule closely enough to know which seminar I would take first." That is more convincing than a feeling about a campus you saw for an hour.
For how much colleges weigh interest itself, see what 46 Common Data Sets say about demonstrated interest. Applying from outside the US? Demonstrated interest for international students covers remote ways to engage, and the why us essay guide for international students covers writing the answer from abroad.
Questions about researching a college for a why us essay
Can I use an AI chatbot to research colleges for a why us essay?
At Cornell and Northwestern, yes for research: Cornell lists "Using AI tools to research specific colleges" as acceptable, and Northwestern lists "Researching colleges and admission terms" as constructive. Other colleges publish different rules, so read each one. Confirm every name the chatbot gives you on the college's own pages.
What if an AI tool named a professor or course I can't find?
Leave it out. If it is not on two college-controlled pages, one of them current, it does not go in the essay.
How many sources do I need for each school fact?
Two pages the college controls, with at least one current for this term or this year. One current official list, such as this year's first-year seminar list, can be enough on its own.
How many facts should I collect per college?
Enough to fill a six-row log and keep two or three after the checks. A longer list does not help, because a 250-word answer has room for about two features.
Does a campus visit help my why us essay?
Only if you learned something you can name. Duke says students "don't need to visit," and a student who never visited can write the same answer from the college's own pages.
Get a free review of your why us essay
Get a free Quick Review of your why us essay against GradPilot's why-us supplement rubric. It reads whether each school feature is described by what it lets you do, whether it ties to something already true of you, and whether your answer completes every question the prompt asks without another school's name. Paste the prompt and its limit above your draft. For a prompt that also asks about your major, the why-major supplement rubric reads that half, and the Common App essay review covers the rest of your list.
More why us guides
- How to write a why us essay: how many features each word limit holds, shown at four lengths.
- Can you reuse a why us essay?: one evidence bank adapted for three colleges.
- Why us essay examples, annotated: five constructed answers at five real prompt shapes.
- Common App essay guides: every college essay guide, by topic and stage.
Sources
Retrieved October 30, 2026, except where noted.
- Duke University Undergraduate Admissions, Resources (no year stated): college websites, "demonstrated knowledge," visiting not required.
- Cornell University Undergraduate Admissions, Receiving Help with Your Application (no year stated): acceptable and unacceptable help, including AI research and AI drafting.
- Northwestern University Undergraduate Admissions, Artificial Intelligence and Admission (updated 9/23/25): constructive and problematic uses of generative AI.
- University of Wisconsin–Madison Admissions, Essays (no cycle year stated): use of AI in admissions essays.
- University of Notre Dame Undergraduate Admissions, Application Overview (redirects to the how-to-apply page; no year stated): AI policy and the Common App Fraud Policy wording.
- Tufts University, Short Answer Questions (Class of 2031): Short Answer Prompt #1, research suggestions and AI note.
- University of Michigan Undergraduate Admissions, First-Year Essay Questions (no cycle year stated; HTTP 403 to scripts, read through a text proxy): Essay 2.
- Brown University Admission, Selecting Degree Programs and Academic Fields (no year stated): concentrations, no quotas.
- National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1, July 2024): definition and risks of confabulation.
Get feedback on your supplemental essay
Paste your draft. The free Quick Review shows what to fix against the prompt's rubric before you submit.