Copy article

Student AI use is turning degrees into a trust problem. Surveillance is the wrong fix

ended 18. March 2026

A new HEPI survey on student use of generative AI lands at the wrong moment for universities: employers are already questioning what a degree proves, and campuses are struggling to separate learning from outsourcing.

The argument is often framed as cheating versus innovation. That misses the real issue. When AI becomes the default shortcut, qualifications start to measure access to tools and prompt craft, not understanding. The losers are not only lecturers. It is students who leave with brittle skills and a CV that cannot survive a technical interview or a first week on the job.

The second-order effect is institutional. Once suspicion becomes normal, everything becomes surveillance: more proctoring, more policing, more conflict. That is expensive, and it corrodes trust between staff and students.

A better response is to change what is assessed. Ask for proof of work, decision logs, and messy intermediate artefacts. Use short oral defences, in-class drafting, and practical tasks where the process matters as much as the final answer. Reward judgement, not fluent paragraphs.

If universities cannot show what a graduate can actually do without a black box, the credential inflation will not stop at essays. It will spread into hiring tests, professional exams, and workplace training.

We'd like your views:

  • What should assessments test in an era where text generation is cheap and instant?
  • Are universities overcorrecting towards surveillance, and will that backfire?
  • How should employers interpret grades if AI assistance is uneven and hard to audit?
  • What does 'acceptable use' look like in practice for students and faculty?
  • Will this push more value into apprenticeships and work-based credentials instead?

2 responses from the Newspage community

Copy all

Star Quote
Copy

Rohit Parmar-Mistry
Founder at Pattrn Data
HEPI's student generative AI survey lands at the worst time for universities: employers already doubt what a degree proves, and AI is now close to universal student kit. Framing it as cheating versus innovation misses the real issue. When AI becomes the default shortcut, qualifications start to measure access to tools and prompt craft, not understanding.

The obvious response is surveillance: more proctoring, more policing, more suspicion. That will backfire. It is expensive, adversarial, and it trains students to optimise around the rules. Better fix is assessment design. QAA has been pointing in this direction for a while: assess the process, not just the final prose.

In our audits, the real pattern is evidence of work: short oral defences, in-class drafting, decision logs, version history, and practical tasks where the messy middle matters. Employers will adapt with hard interviews and work tests. Universities should get there first, and stop pretending detection is a strategy.
Copy

Editor
Kate Underwood
Founder & Chief People Strategist at Kate Underwood HR and Training
Will this push more value into apprenticeships and work-based credentials? Absolutely, and it's already happening. Employers are tired of graduates who can write a perfect essay but can't solve a real problem or handle a difficult conversation. When you can't tell if the degree proves competence or just access to ChatGPT, the credential becomes worthless. Apprenticeships and on-the-job training show what someone can actually do, not what an AI can produce on their behalf. Small businesses have been saying this for years: give us someone who can think, adapt, and learn on the job, not someone with a 2:1 and no practical skills. Universities are pricing themselves out of relevance while producing graduates who aren't work-ready. If a degree can't prove capability anymore, employers will stop paying for it. The shift is already underway. Apprenticeships, vocational training, and demonstrable skills will win because they can't be faked by a chatbot.