Recent events at Texas A&M University highlights the potential problems caused by using AI in education settings. A tutor at the University who accused his students of using ChatGPT to write their essays; he used ChatGPT himself to copy the student essays into the system, asking it to identify if it [ChatGPT] had written the assignments, and failing those which it identified as matches (despite ChatGPT not being an AI detector). This case did not appear to lead to any drastic implications for the students or the tutor, but it does present potential areas for disputes to arise. 

One of the key questions is of academic integrity. If students are using AI to complete their assignments, does this constitute cheating? The answer to this will ultimately be determined by an institution’s policies. No doubt UK University will have well-established policies on specific matters of academic integrity, such as plagiarism or falsification, all of which form part of the contractual relationship between student and institution. But do these policies include provisions which address the wide stream of AI models which are rapidly emerging? Is there an outright ban? Is conditional use permitted? Is there silence on the matter all together? Ambiguity in these areas certainly poses the risk of disputes where penalties are imposed for the use of AI.

The obvious follow-on issue is that even if there is a clearly defined policy on AI models, how would breaches be enforced? The gold-standard plagiarism detectors ascertain their reliability by being able to pinpoint which extracts had been plagiarised, and more importantly where from, providing the specific reference of the original work. AI disrupts this process as the content it produces is entirely new. Whilst some AI-detection models are beginning to surface, their accuracy is undermined by the risk of false positives being made, which could generate a new wave of disputes from students who are wrongly penalised.