UNC Chapel Hill Ends AI Scoring of Undergraduate Admissions Essays

What happened

The University of North Carolina at Chapel Hill has announced that it will stop using artificial intelligence to rate undergraduate admissions essays, according to a report from GovTech.

The available reporting is brief. Details such as the timing of the change, the specific tool involved and the university's stated reasoning should be confirmed against the original GovTech article before they are repeated. What is clear is the direction of travel: a public university is withdrawing AI from a step that directly shapes how applicants are assessed.

Why it matters

For public institutions, using AI in a high-stakes judgement such as admissions is as much a question of trust as of technical capability. Essays are written by people and, in most processes, read by people who are accountable for the outcome. When an automated score enters that chain, applicants need to understand what is being measured and how a human can review or challenge the result.

The decision also shows that AI adoption in government-linked services is not a one-way ratchet. Organisations can pilot, review and withdraw tools as evidence and expectations change. Leaders planning similar deployments should map where an AI output is used, who can override it, and how any change is explained to the people it affects.

The Renascence take

Most of the commentary will focus on whether AI can judge writing well. The more useful question is whether applicants can see the logic behind a decision that shapes their future.

An essay score that looks objective can make a process less legible, not more, and legibility is what sustains trust. Perceived fairness usually drives acceptance more than accuracy does, so a technically sound model can still damage the experience if nobody can explain it. Customer-obsessed operators should audit every point where an AI output feeds a decision, publish a plain-language account of how it is used, and keep a named person accountable for the final call. If a tool cannot survive that scrutiny, withdrawing it is a service decision, not a failure of innovation.