AI in Research: a sceptic’s reflections

As a second-year PGR conducting research on the use of digital technologies and predictive policing in the work of frontline police, it is ironic that I tend to avoid using AI myself. Indeed, I would place myself in the ‘sceptic’ end of the ‘trust in AI spectrum’ – particularly when it comes to academic research.

But throughout my PhD journey so far, one skill I’ve been keen to improve is my writing style, specifically my ability to write in plain English to a general audience. This is something I have struggled with, often overcomplicating my sentences and using unnecessary jargon without ‘getting to the point’. My supervisors have been helpful in this respect, pointing out ways I can write more concisely, with greater clarity and confidence. At the same time, I was conscious that such work on their part was taking up time better placed on feedback relating to the actual content of my writing.

Curious to know how other researchers felt about the use of AI in their work, I attended the launch event of the newly established Artificial Intelligence Researcher Development Network Plus (AI.RDN+) held at Aston University in January of this year.

Led by Aston and the University of Leeds, the network represents universities across the Midlands and Yorkshire and aims at building knowledge on the implications of AI for doctoral research. Though I maintain my scepticism, I came away from the event feeling enlightened as I learnt about some of the opportunities of using AI for my writing needs. This included employing AI as a tool for making editorial suggestions – like having a second (or third) supervisor. The idea is that AI helps to increase efficiency by taking on some of the more mundane tasks which would otherwise take up supervisor’s time. However, the event also cautioned against using AI to reach conclusions. AI can offer guidance and support reasoning but cannot replace critical (and creative) thinking. Developing skills such as academic writing takes time and should ultimately come down to the PGR.

I consolidated some of these insights by attending the AI for academic writing for PGRs workshop offered by Libraries and Learning Resources. The workshop explored the benefits and risks of using GenAI for academic writing, and how to use effective prompts to get the most out of GenAI as well as consideration for data integrity and copyright issues. The workshop also taught me about the types of AI, how GenAI actually works, what GenAI tools are available, and which ones are recommended. Moreover, the workshop covered the University policy on using AI, something all users should familiarise themselves with. For those unsure about how AI can be used responsibly as a tool for assisting academic writing, I highly recommend attending one of these workshops.

Overall, both the networking event and workshop together were thoroughly insightful for developing my understanding of how AI can be used responsibly in research and as a tool for academic writing. At the same time, and thinking particularly about supervisor relationships, they also made the important point that AI cannot replace the human side which is, at least in my view, a fundamental aspect of the PGR experience. On that note, I will end with a quote from Professor Tom Stoneham (University of York): “Sometimes being responsible means not using AI at all”.

And on one more final note – I would also like to state that, perhaps expectedly, no AI was used in the writing of this blog…