Watch this space…
Over the last few weeks and months I’ve been working with colleagues across my institute on drafting specific guidance for doctoral students around the use of Generative AI in their studies.
It had become increasingly clear that the initial student facing guidance was inadequate, focussed as it was on taught students. This itself is perhaps not surprising – the initial risk of academic misconduct was much higher, and policies needed to be drafted quickly in order to provide clear guidance.
After all, what’s the problem – at the time of writing, the LLMs underpinning most GenAI tools are not capable of delivering a whole thesis, and the student still has to undertake a viva examination defending the manuscript. Many of the features of the doctoral process are being taken up by faculty teaching undergraduate students, including real-time questioning.
The main concern is the nuances involved – can a student upload their own thesis and generate viva questions to practice with? If a student has used GenAI to support the code development, does that need to be declared? To what extent can GenAI support the writing of the thesis?
My initial suggestion was that we keep the guidance short – that students remain responsible for the intellectual contribution of their doctoral research, and that any use of a GenAI tool must be declared. We could then leave it to standard processes for any concerns to be raised.
Things are neve that easy and the guidance has steadily grown, as have the number of people involved – consulting with student representatives, checking across disciplinary differences, consulting copyright experts, engaging the IP team, validating the guidance with the AI leads, oversight from the Graduate School, considering the need for supervisory training, and ensuring disability support were happy with the guidance.
This process – and the final guidance – is almost the least interesting part though. The essence remains the same; know what you are doing, and remain in control of your own research. GenAI is going to fundamentally change some academic-related activities – I use it all the time to support brainstorming, draft article structuring, experiment with assessment questions – it’s integration into research practices remains in the early stages.
Beyond support in writing, some tools are using GenAI to support elements of literature review. Personally I remain stuck in my ways, and have developed my own techniques for skimming papers, but undoubtedly these tools may prove useful to researchers dealing with the ever-increasing avalanche of literature constantly being published. GenAI may be good at identifying areas of under-considered research interest, checking experiment protocols, or generating meaningful interview questions.
Even if the tools are perfect in each of these scenarios, I believe my initial response will remain meaningful students remain responsible for the intellectual contribution of their doctoral research, and that any use of a GenAI tool must be declared. While we remain in the hype bubble, it’s too early to tell how fundamentally the activity of research will change – but the printing press, the calculator, the Internet and Wikipedia have all come and become integrated into our every-day practices, while intellectual rigour remaining with the researcher. We will have to watch this space to ensure the same remains true with GenAI.