AI is flattening the college essay. Here is what that does to your test score.
As AI makes application essays sound more alike, a proctored SAT or ACT score carries more independent signal. That is not proof testing is fair, but it changes whether to submit one.
The short version
AI writing tools are making application essays converge. They increasingly sound alike and say less. Analyses of large essay samples find that AI-written essays are markedly more similar to one another than human-written essays are, and a 2026 study of more than 370,000 applicants found that after ChatGPT, essays used more varied words while expressing less varied ideas. As the essay becomes a weaker, noisier signal, an independently produced, proctored test score carries relatively more information. That does not make testing unbiased. It changes what each part of your application is worth.
What the research shows
- Essays are converging. Large-sample analyses find AI-written essays are much more similar to one another than human essays are: two randomly chosen AI essays read more alike than two randomly chosen human ones.
- More words, fewer ideas. A study of more than 370,000 applicants found that after ChatGPT, essays gained surface variety in vocabulary while losing variety in their underlying ideas, both within an essay and across applicants.
- The flattening is uneven. Researchers found the convergence was often strongest among lower-income and rejected applicants, and that estimated AI use rose fastest among lower-income applicants. A caveat that matters: these studies estimate similarity to AI writing, not proof that any individual used AI.
- Colleges noticed. Some reduced or dropped supplemental essays for the 2026-27 cycle; Tulane pointed to generative AI, while others cited reader workload.
The honest read
Two things are true at the same time:
- As essays homogenize, a proctored score is one of the few remaining signals that is independently produced and hard to outsource, so it gains value relative to the essay. (Selective schools are reinstating test requirements too, though they cite their own data on how well scores predict college performance, not essay homogenization, as the reason.)
- That does not mean the score is unbiased or should dominate. Scores carry their own socioeconomic signal, including the retake gap. The real lesson is that every part of an application has relative, not fixed, value: when one signal gets noisier, the others matter more.
What it means for you
- A strong score reasonably carries more relative weight than it did a few years ago, especially at a test-optional school, because the essay beside it is a noisier signal than it used to be.
- Do not outsource your essay to keep up. One study found heavier estimated AI use associated with worse predicted outcomes, and homogenized essays are exactly what is losing value. A specific, genuinely human essay stands out more now, not less.
- Let your score do verifiable work. If your essay is going to look like everyone else's, your score is where you differentiate on ground a reader can trust.
The bottom line
AI is making the essay say less, which quietly raises the value of a real proctored score and of a genuinely human essay. Lean into both, and do not let a chatbot flatten the one part of your application that is supposed to sound like you.
Practice the pattern, not just the question. Every drill and full-length SAT and ACT test here shows you which trap you fell for after each question. Free, no sign-up to start.
Try a real question