With unclear guidance, students are learning to manage AI quietly – and universities are none the wiser.

Since the release of ChatGPT in 2022, generative AI has become central to university study habits. From using AI as a coding tool or even as a stand-in to avoid doing work entirely, students across the UK have made it a fixture of academic life. 

But with institutions scrambling to keep up, clear policies around AI usage are rare and instruction on how to effectively use the technology is non-existent. Experts warn that a serious disparity in AI literacy may be taking root.

In March, a Higher Education Policy Institute (HEPI) survey revealed that generative AI use among  undergraduates is near universal, with some 95 per cent of students using AI in at least one way and 94 per cent using it to help with assessed work. Universities no longer need to ask whether AI is a part of students’ study habits. 

The crucial question, says Dr Phil Anthony, head of AI at the University of Kent (UoK), is whether universities are teaching students to use the tech appropriately or whether they are allowing them to outsource their thinking completely. Failing to do the latter risks creating a “hidden curriculum” around AI, he argues.   

“If AI is only discussed through the lens of misconduct, students may become reluctant to disclose how they are using it, even when their use is sensible or exploratory,” he explains. “That can push AI use underground. Students may turn to unsupported tools, rely on advice from peers or social media, or avoid asking questions because they are worried about getting into trouble.”

“If AI is only discussed through the lens of misconduct, students may become reluctant to disclose how they are using it, even when their use is sensible or exploratory”

Alice Houghton is an undergraduate student studying geography at the University of Manchester (UoM). She says that she tries not to use AI for anything more than checking grammar because some lecturers are opposed to its use. 

“UoM does permit the use of AI with moderation,” she says. “But the guidelines aren’t really enforced, and this means that it usually depends on the lecturer. I have several who are totally against it and so I find it easier to not use AI as much as possible just to be safe.”

Guidelines such as those implemented by UoM are a start, but, as they are not robust academic policy, have little day-to-day impact on students’ AI usage. As Alice notes, it also leads to individual lecturers dictating students’ study habits.

Anthony notes that this is likely to lead to an equality issue, with students not provided a level playing field or equal footing on which to develop their AI literacy.

“Students who are more confident or already familiar with AI may continue to use it in ways that give them an advantage, while others may avoid it completely or use it less safely,” he says. “Silence around AI does not make AI use disappear. It simply makes it less visible, less supported and potentially less fair.”

Max Burrows is a postgraduate student studying robotics at King’s College London. Because his course focuses on technological advancements, the use and study of AI is encouraged.

“I use AI with basically all of my work,” he explains. “I’d say in most of my specialisms it’s a necessity because the course is preparing you for careers that are directly in AI or now revolve around AI.”

His confidence with the technology extends beyond his current course.

“In my engineering undergraduate I was taught how to engineer prompts in pretty advanced ways and now in my Master’s, a lot of my work is focused on machine learning … I’m confident I’ll get a job either within the AI sector or related to AI.”

That imbalance matters more as working with AI becomes a baseline expectation from employers. In a 2026 report, AI training platform DataCamp found that business leaders see working with AI as the most important skill in the workplace. 

For Anthony, that means universities need to reframe how they think about AI entirely. “Employers are unlikely to need graduates who simply outsource thinking to AI. They will need graduates who can use AI to support good work while still bringing disciplinary knowledge, ethical judgement and accountability,” he says.

“Universities therefore have a responsibility to prepare students for that reality, rather than treating AI only as a threat to assessment.”

So how exactly can universities accomplish this? Anthony argues that the key is to develop context-specific policies that are connected to learning outcomes, as opposed to broad institutional statements. “It should explain what is permitted, restricted or not allowed in plain language,” he explains, adding that it should also explain the rationale behind it so students won’t see it as an “arbitrary rule”. 

Finally, it should “tell students how to acknowledge or discuss their use of AI where that is appropriate”.

Back in Manchester, Houghton is not convinced that change is coming any time soon. “Most of my friends that do use AI a lot don’t really use it in an ‘efficient way’,” she says. “Most of them in essay-based subjects just use it to write up drafts or to find sources… I do worry sometimes about how important AI is becoming and if the skills I’m prioritising will be taken over by AI-based ones.”

As the technology becomes increasingly ubiquitous in academia and in the workplace, students will find themselves becoming increasingly drawn to it, whether their specialties lie in STEM or the humanities. Universities must begin to ask – and quickly answer – the question of whether they teaching students to use it to the best of their abilities.

Featured Image Caption: Hand types in search bar using AI ​​Prompt. AI powered search engine. Credit: Adobe Stock