AI is now part of everyday student work, while universities are still working out how to assess it fairly. A University Times opinion article says 94% of UK undergraduates surveyed by HEPI used AI for assessed work in 2026, and raises a harder question: can institutions reliably tell when a student’s writing is their own?
Watch Desk analysis
What happened
The article, published on 13 November, describes Trinity College Dublin’s approach: its central guidance leaves individual schools to set their own rules, which can range from citation requirements to bans. It says students may therefore face different standards across courses. The article also notes that Trinity’s Blackboard modules use Turnitin, which provides both similarity reports and an AI score.
The author cites Stanford researchers’ finding that seven AI detectors labelled 61.22% of essays by non-native English speakers as AI-generated. The article also reports that Turnitin’s own figures put 14.8% of English-language essays submitted between October 2025 and February 2026 in its category of at least 80% AI-written, compared with 3.3% in 2023. Those are claims and figures presented by the article, not a measure of how often students actually used AI.
Why it matters
A detector score can influence how a lecturer views a submission, but the figures cited here underline why it should not be treated as proof on its own. If clear, careful writing is more likely to be flagged, students writing in a second language may face an uneven burden of suspicion.
The practical challenge is bigger than catching copied text. Students need to know what their course permits, and universities need assessment methods that test learning without asking software to settle authorship by itself.
Our read
The useful starting point is clarity: publish course-level rules where students can find them, and do not let an automated score stand in for a conversation or evidence. AI use is already widespread in the HEPI survey cited by the article. Pretending it can be handled with one detector and a stern note in the handbook looks rather like bringing a smoke alarm to a philosophy seminar.
What to watch
- Whether universities make AI rules consistent and visible across courses.
- Whether institutions disclose how detector scores are used in academic-integrity decisions.
- Whether future research finds detectors perform reliably across different kinds of student writing.
Discussion spark: Should universities rely on AI-detector scores when investigating coursework, or should they have to show independent evidence before challenging a student?
Sources and evidence
- AI Is Training To Be A Better Student Than You Are – The University Times (14 November 2026, 00:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.