AI use in UK universities varies between institutions, so there is no single rule covering every student. You need to check your own university and module policies, understand what forms of AI use are permitted, keep an audit trail of your work, and know how AI use must be declared or referenced.

What is the general position on AI use across UK universities?

The Russell Group has developed principles on the use of generative AI, stating that its universities want AI tools to be used for the benefit of students and staff while supporting students to use them appropriately and effectively.

Most universities have now produced some form of policy on AI use, but the specific rules differ between institutions.

For example, the University of Portsmouth states that AI-generated content must not be included in assessment submissions unless this is specified in the assessment brief or approved by the module lecturer. Its guidance also covers text editing and proofreading, stating that students can use AI to suggest edits and grammar changes but must edit the work themselves and acknowledge AI use.

The University of Glasgow similarly states that submitted work must represent the student's own effort and strongly discourages using AI tools to check writing. The University of Manchester, on the other hand, allows some uses such as spell checking, idea generation and revising understanding, while requiring transparency about AI use.

The important point is that you need to check your own university's policy.

How is AI use detected in university assessments?

AI detection tools have been developed to identify possible AI-generated writing, and Turnitin now includes AI detection. Its system analyses sections of a submission and produces a score representing the probability that the writing was generated by AI.

AI detection relies partly on patterns in language generation. ChatGPT generates text by selecting probable next words, creating patterns of consistency that detection systems can analyse.

However, AI detection tools are not always used. At the University of Portsmouth, for example, tools are not used because of concerns about how they work and how their results should be interpreted. Academics may instead look for signs such as changes in writing style, lack of original insights, inconsistencies, fabricated references, sudden changes in tone, or differences in the depth of analysis.

If AI use is suspected, whether through detection software or academic judgement, the matter may be referred to an academic offence panel. Students may then be asked for evidence such as previous drafts, working documents or raw background research to demonstrate how the work was produced.

What should you look out for when using AI?

Firstly, it is your responsibility to understand your university's policy. Policies can differ considerably, and the information may be spread across several documents, including academic standards, academic offences and marking policies.

Be clear about:

• what AI use is allowed

• under what circumstances it is allowed

• whether you need permission from your tutor

• how AI use should be declared or referenced

Secondly, keep an organised folder containing your drafts and background research. False positives can occur with AI detection software, so maintaining a good audit trail of your work can help demonstrate how your submission developed.

Finally, learn how your university expects you to cite and declare AI use. Some universities require students to declare AI use on the cover page and within the submission, including where and how AI was used.

Should students use AI in their university work?

Learning how AI can be used in research can be a valuable skill, particularly as companies start to expect AI literacy from graduates. However, you need to be careful about whether your university permits the particular use you have in mind.

You should also consider what skills you are developing through your degree. AI should not replace opportunities to develop skills such as writing, proofreading and analysing data.

The fact that an AI tool can perform a task does not automatically mean that using it for that task is appropriate for your assessment. Your university's policy, your module requirements and the skills you are expected to develop all matter.

The safest approach is to understand the rules before using AI and keep clear records of your own work throughout the research and writing process.