AI in Universities: Students Are Confident, Staff Are Not

Students tend to assume their university will catch anyone who misuses AI. Lecturers, on the whole, know it is not that simple. That mismatch is one of the biggest talking points around AI in universities, and it matters for the fairness of every degree.

What students believe about AI in universities

Take-up is rapid. Globally, 54% of prospective students say they plan to use tools like ChatGPT simply to choose where to study. In one 2025 student survey, three quarters (76%) said their institution would notice AI use in assessed work, and 80% said their institution’s AI policy was clear.

What staff are saying

A global survey of faculty tells a different story:

• Only 6% fully agree that their institution has given them enough resources to build AI literacy, so 94% feel under-prepared or under-supported
• 83% worry about students’ ability to judge AI-generated material critically
• 54% think their current assessment methods are no longer adequate, and 13% want an urgent, complete overhaul

Confidence is growing on one side of the classroom while capability is lagging on the other.

Why the gap is risky

Fair enforcement depends on consistency, and that is hard to deliver when most staff lack training. Identical behaviour could lead to different outcomes depending on the course, which weakens trust in the whole system.

There is a thinking problem too. Analysis and critical thought are where generative AI is weakest, so assessments that only ask for summaries are easy to game. Meanwhile, if staff do not know what evidence to look for, such as factual errors, inconsistent tone or missing sources, there is a danger of wrongful accusations. Every disputed case then adds pressure to teams that are already stretched.

Ben Dickinson, Senior Analyst at BoostMyGrade, suggests treating AI detectors as leads rather than verdicts. He recommends basing decisions on draft histories, short oral discussions and versioned work. Until staff are properly supported and assessments redesigned, he expects enforcement to stay uneven.

What institutions can do

The issue is less a looming disaster than a coordination challenge. Suggested steps include:

• Fund training. Put real resources into AI literacy for staff and create consistent, system-wide policies.
• Redesign assessments. Reward analysis and critical thinking, and use in-class tasks where outside AI help is tempting.
• Set clear evidence standards. Teach staff to spot inaccuracies, work that ignores course content and fake or dead-end links.
• Be transparent. Ask students to cite any AI use and explain how much they relied on it, with fair process for everyone.

What this means for students

For anyone studying, the safest approach is to read your own institution’s rules closely, because policies and practice can differ from one course to the next. Keep your drafts and notes as you go, since they show how your work developed, and cite any AI help you use.

It is also worth remembering that critical thinking is exactly what assessments are shifting towards. Relying on AI for a summary may not be enough when the task asks for your own analysis, and staff are increasingly looking for signs such as errors, missing course content and links that lead nowhere.

The risk of doing nothing is a loss of faith in the value of the degree itself. Treating AI literacy as essential infrastructure, not an optional cost, looks like the sensible way forward.