‘Near universal’ student AI use leading to ‘polarised views’

The rapid uptake of generative artificial intelligence among UK higher education students had led to a “polarised landscape” with little agreement over whether it is helping or hindering learning and well-being, according to a new report.

AI use has become “almost universal” among undergraduates in just three years, the Higher Education Policy Institute (Hepi) found after polling students for the third time since the launch of ChatGPT propelled large language models into the mainstream.

Ninety-five per cent of students now report using AI in some way, up from 92 per cent a year ago and 66 per cent in 2024.

Generating text is still the most popular use, though this has dipped slightly in the past year from 64 per cent to 56 per cent.

However, the proportion of students directly including AI-generated text in assessed work has risen to 12 per cent, up from 8 per cent in 2025 and 3 per cent in 2024.  

Summarising textbooks, note-taking, generating images and video and data analysis are among the uses that have increased in popularity in the same period.

Almost half (49 per cent) of students believe AI has improved their student experience, with saving time, improving understanding and providing instant support among the advantages.

But 16 per cent feel AI has worsened their experience, citing concerns about fairness, skill erosion, social isolation and future employment.

This ambiguity over whether AI is a good thing is mirrored in many of the other findings of the report.

In comments submitted to the poll, one student praised the tools for summarising dense readings and saving hours of “tedious work and letting me focus on critical analysis and deeper understanding” whereas another simply said “I’m not using my brain at all”.

A similar proportion of students (20 per cent) felt that AI made them feel lonelier as those who said it made them feel less lonely (21 per cent) with some saying they use the tools for companionship and advice.

A further divide was seen over the question of institutional support, with 37 per cent saying they feel encouraged to use AI by their institution, and 36 per cent saying the opposite.

While the findings show that institutions are still broadly playing catch-up in their response to AI, assessment practices have changed dramatically – with some students complaining this had made things more difficult and others expressing anxiety about false accusations of misconduct.

The report recommends universities consider changing their curricula to more explicitly teach AI knowledge and skills across all subjects.

There is still a need to publish clear, accessible and assessment-specific guidance on AI use, it adds.

Institutions should also “ensure that the AI tools necessary or advantageous for course content or assessment are accessible to all students”, amid continued concerns that only the wealthy are able to take out subscriptions to the more premium services.

tom.williams@timeshighereducation.com