Privileged students ‘more likely to use ChatGPT’ on assessments

Universities should consider funding subscriptions so students have equal access to the best generative artificial intelligence tools, amid signs of an “emerging digital divide” that has seen those from privileged backgrounds adopt the latest technologies more quickly, a report says.

Polling of 1,250 UK undergraduate students published by the Higher Education Policy Institute (Hepi) on 1 February finds that more than half (53 per cent) of them are already using tools such as ChatGPT to help with their assessments.

Uptake has been higher among those from more privileged backgrounds, with 58 per cent saying they use it, compared with 51 per cent of those from the least privileged backgrounds.

Despite the high usage of the tools, only a fraction (13 per cent) of the students say they have used AI to generate text for an assessment, and a smaller proportion (5 per cent) say they have then submitted this text without editing it.

Instead, most students say they use AI tools to explain concepts, suggest research ideas, summarise articles or assist with grammar, translation and essay structure.

“There has understandably been a lot of concern about widespread cheating, but this report shows that doesn’t seem to be happening and instead students are using it in ways that I think most people would view as acceptable,” said Josh Freeman, Hepi’s policy manager and author of the paper.

“Whilst the survey shows there are a lot of students that are real experts in AI, there are also plenty of students who don’t really know how to use it and who have never clicked on it,” he added.

There were not yet “enormous differences” between the most and least privileged in terms of AI use, Mr Freeman said, “but there seems to be a gap that appears to be widening”.

The survey also finds that Asian students are the most likely ethnic group to use AI and that men use it more than women.

Companies have tended to introduce fees for access to the more advanced versions of their products, which may explain some of the discrepancies, Mr Freeman said.

His paper recommends that “when generative AI tools are beneficial for learning, efforts should be taken to equalise access to them”.

If students are permitted or encouraged to use paid-for AI tools, it adds, “institutions should consider funding subscriptions to these tools, as many institutions already do for other digital tools”.

A majority of students in the survey feel their institution has a clear policy on AI use. However, they are less positive about the support they had received to use AI, which Mr Freeman said was reflective of institutions being quick to outline what students could and could not do with the technology, but less adept at teaching students how to use it in a helpful way.

tom.williams@timeshighereducation.com