Galley 18_1_1618 Betteney, et al. Opinion Piece Compass: Journal of Learning and Teaching, Vol 18, No 1, 2025 13 Generative artificial intelligence and the doctoral student: procedural and ethical considerations. Mark Betteney, Michelle Cox, Xiaonan Duan, John Kpukumu, Meena Malhi, Della Mulhare, Mutiat Olaifa, Bing Yin University of Greenwich Abstract This opinion piece reports the discussions of a group of seven EdD (Doctor of Education) students and their tutor (lead author), who together have co-authored this paper. The piece identifies our diverse attitudes, anxieties, aspirations or apprehensions regarding the use of generative artificial intelligence (GenAI). The discussion evolved to explore six main areas: academic literacy; ethical use of GenAI; the need for originality; the consistency of attitudes held by assessors and supervisors; the specificity of university guidance; and the viva as an assessment tool. The extent to which aspects of the discussion resonated with examples of published work was also considered. Keywords: generative artificial intelligence, doctorate, ethics Opinion Piece Compass: Journal of Learning and Teaching, Vol 18, No 1, 2025 14 Introduction The University of Greenwich gives separate guidance to staff and students regarding the use of generative artificial intelligence (GenAI). For staff, the guidance (UoG, 2025a) covers GenAI for teaching, research and administration. For students (UoG, 2025b), the guidance is predominantly focused on academic literacy and conduct, identifying some of the ways by which GenAI might cause users to offend against established academic and ethical standards. The university overtly encourages students to take advantage of GenAI to support their learning and its approved GenAI tool is Microsoft Copilot. Since doctoral students undertake research, they need to be cognisant of both sets of guidance. This opinion piece reports on our discussion, which revealed that the co-authors – seven EdD students and their tutor1 – had diverse attitudes to GenAI. Before the discussion, three students distrusted all artificial intelligence (AI) for academic purposes, two were agnostic and two were positive about the benefits of GenAI for their studies. The differences in attitudes did not correlate with the different ages of the students. Six areas of interest emerged from our discussion, which will be presented here in turn. Consensus was achieved only in two areas, but our discussion emphasised the challenges presented by GenAI to universities and students alike. Academic literacy Our discussion group’s views about GenAi in the context of academic literacy were conflicted: the students generally felt that to write some or all of a thesis text with Copilot was morally and ethically wrong, yet each acknowledged, without a qualm, using AI tools such as spellchecks; they also considered proofreading tools for improving language to be within the bounds of acceptability. We found it impossible to delineate where acceptable use of GenAI and AI starts and stops within the field of academic literacy, beyond the university’s requirement for clear referencing and acknowledgement of AI and GenAI use (UoG, 2025c). The difference in functionality between AI and GenAI was also impossible for us to identify discretely and it was felt that perhaps herein lies the nub of the challenge that all universities, teachers and students are having to face. Our discussion chimed with the expressed attitudes of Ali et al. (2025) and Tuncer Bayramli and O'Shea (2024) who each acknowledge this tension, and who argue that AI can not only act as an augmentation to academic literacy skills, but can also serve to enable accessible and equitable scholarly practice, as well as to facilitate positive change in scholarly discourse if it is embraced for use as a recognised tool, not exclusively considered as an avenue for dishonesty. Ethical use of GenAI There was some disagreement amongst us about the ethics of GenAI use in doctoral studies. For example, before writing a literature review, is it acceptable to ask Copilot to identify the 1 No-third party data was collected for this opinion piece. The views expressed constitute only the opinions and discussions of the co-authors and thus no ethical approval was needed. Opinion Piece Compass: Journal of Learning and Teaching, Vol 18, No 1, 2025 15 main themes that would appear within the literature underpinning a given research question? Or, given that no one objects to students using Statistical Package for the Social Sciences (SPSS) to analyse numeric data, or NVivo for qualitative data, is it ethical to use a GenAI tool for data analysis? Attitudes among us were conflicted on this question because GenAI draws on algorithms and other text data, and future GenAI use will in turn be informed by the data students provide to the GenAI tool and also by the analysis it has made of the data; consequently we were uncertain whether the use of GenAI to analyse data would constitute a breach of data protection. Cunningham (2024) identifies key risks for data analysed through GenAI, of which four are pertinent to the work of Doctor of Education (EdD) students. The risks are: lack of transparency (it is hard to inform participants exactly how their data is kept safe or is subsequently processed if AI has been used in analysis); secondary use (participants will have no say about which subsequent studies their data may be used for); data subject rights (it is not possible for AI models to respond to requests from participants to withdraw, rectify or erase data); and lawfulness and accuracy (third party data that has been scraped from the internet cannot be said to be lawfully collected). The need for originality To be successful, doctoral students need to claim and demonstrate originality. GenAI cannot provide that, being a predictive tool based upon statistical probability. Extensive use of GenAI, like the examples in the section above, would be an impediment to developing originality in a thesis. In this at least, our group was unanimous. The consistency of attitudes held by assessors and supervisors Just as there is inconsistency within the attitudes of students towards GenAI, there must also be differences in the degree of tolerance or confidence of tutors towards it. Some tutors might be ‘old-school’ or even antagonistic; some might be GenAI-innovative. If student and supervisor held very different attitudes, this could fundamentally jeopardise the relationship. Students in our discussion felt there could also be variety in the responses of tutors on seeing a declaration of AI or GenAI use in a submission, which could possibly advantage or disadvantage students when being assessed, depending on how the assessor felt about GenAI. The specificity of university guidance The use of GenAI in academic processes is in its infancy, and we students understood that the university guidance is still evolving. Although the university encourages students to use AI, we felt that we (and potentially all current students at all levels) were, if not guinea pigs, then at least pioneers in its use. We considered that the university’s guidance is at present rather non-specific and it does not clearly identify what is allowed and what is not. This leads to student anxiety both amongst those cautious about GenAI and, perhaps especially, those who are enthusiastic. Opinion Piece Compass: Journal of Learning and Teaching, Vol 18, No 1, 2025 16 The viva as a doctoral assessment tool Doctoral students are assessed through a viva and here again there was consensus. The thesis must be defended and no amount of GenAI use will help a student on the day of the viva. Therefore, GenAI use in a thesis should be restricted to where its use is compelling or natural, and ethical. We wondered whether the essay as an assessment tool has had its day, grounded as it is in yesterday’s technologies. It was felt that the viva could become common even within undergraduate courses, as it is an AI-resistant assessment tool, and many in our discussion group would have welcomed that when they were undergraduates. Moorhouse et al. (2023) identify the need for universities not only to publish guidelines about AI use (which most now have done), but also to consider assessment design in order to protect students from falling foul of the standards of academic integrity and slipping into dishonesty. If universities develop AI-resistant assessment tools, they will not only reduce academic misconduct problems; they will also pave the way for GenAI use to become a celebrated digital literacy tool, rather than a threat to academic integrity. Conclusion The EdD students who co-authored this piece began with very diverse views and degrees of confidence about what AI and GenAI were and whether they were a ‘good thing’. Some students softened their positions, others confirmed theirs. Before the discussion, the views of the most cautious students had been formed from ignorance and they had actively avoided GenAI for fear of penalties. Those students went on to express at least a willingness to explore what AI and GenAI can ethically and morally do for them as doctoral students. The University of Greenwich is being proactive in its consideration of GenAI and it is consulting heavily with staff through surveys and special interest groups about their experience of, confidence about and reaction to GenAI use when identified or discovered in students’ formative and summative assessments. Policy development and analyses are being undertaken at strategic levels within the university and disseminated through multiple communication channels, but, in our discussion, we acknowledged that technology is changing faster that policy can. Consensus among us was that, however enthusiastic or proactive a student may feel about GenAI, caution is necessary in its use as a study tool, at least until after our doctorates have been awarded. Reference list Ali, M.Y. and Richardson, J. (2025) ‘AI literacy guidelines and policies for academic libraries: A scoping review’. IFLA Journal. Available at: https://doi.org/10.1177/03400352251321192 (Accessed: 24 June 2025). Opinion Piece Compass: Journal of Learning and Teaching, Vol 18, No 1, 2025 17 Cunningham, L. (2024) ‘The Data Protection Implications of AI: What does your organisation need to know?’ CarsonMcDowell. Available at: https://carson-mcdowell.com/news- insights/insights/the-data-protection-implications-of-ai-what-does-your-organisation-need-to- know (Accessed: 24 June 2025). Moorhouse, B.L., Yeo, M.A. and Wan, Y. (2023) ‘Generative AI tools and assessment: Guidelines of the world's top-ranking universities.’ Computers and Education Open, 5 (1001501), 1-10, Available at: https://doi.org/10.1016/j.caeo.2023.100151 (Accessed: 24 June 2025). Tuncer Bayramli, N. and O'Shea, P. (2024) ‘AI Research tools in Emerging Technologies: their impact on digital literacy and scholarly communications in academic libraries’ EDULEARN24 Proceedings, 1-6. Available at: https://library.iated.org/view/TUNCERBAYRAMLI2024AIR (Accessed: 15 July 2025). University of Greenwich (2025a) AI for research. Available at: https://www.gre.ac.uk/ai- guidance/research (Accessed: 24 June 2025). University of Greenwich (2025b) AI Guidance for Students. Available at: https://www.gre.ac.uk/ai-guidance (Accessed: 24 June 2025). University of Greenwich (2025c) Referencing Copilot and other generative artificial intelligence (AI). Available at: https://www.gre.ac.uk/__data/assets/pdf_file/0032/366179/referencing-for- generative-ai-1.pdf (Accessed: 24 June 2025).