Aston University and the University of Leeds have been awarded £3.4 million to build the UK’s first national network mapping how publicly available generative AI tools are used in doctoral research—and to produce the practical guidance universities desperately need. The Artificial Intelligence Researcher Development Network Plus (AI.RDN+) will run for four years, funded through the Research England Development Fund, and will create a living portal of training, procurement templates, and policy recommendations tailored to the unique pressures of PhD supervision, examination, and research integrity.

Professor Phil Mizen of Aston University leads the project, with Dr Hosam Al-Samarraie and Professor Arunangsu Chatterjee from Leeds as academic leads. The network draws on the eight research-intensive universities of Midlands Innovation and the twelve institutions of the Yorkshire Universities consortium, giving it direct reach across twenty institutions. Sector bodies—Jisc, Vitae, the UK Council for Graduate Education, and the National Centre for Universities and Business—will help channel outputs directly into national researcher development infrastructure.

“The Artificial Intelligence Researcher Development Network Plus will provide detailed knowledge of the uptake and impact of publicly available AI tools across the doctoral ecosystem and use this to co-create much-needed information, resources and professional and skills training opportunities,” Mizen said. “Our project is a unique opportunity to build knowledge and capture innovation, and to use this to build the resources needed for the ethical and responsible use of AI in doctoral research.”

Chatterjee added: “By working alongside Aston and partners across the Midlands and Yorkshire, we can bring together complementary expertise and perspectives. Together, through this network, we can build shared resources and approaches that ensure AI adoption in doctoral research is innovative, ethical and delivers real benefit for researchers and society.”

Why doctoral research demands its own AI rulebook

Most institutional AI policies have been drafted for undergraduate coursework, where concerns about plagiarism dominate. Doctoral study operates on a different plane entirely: projects run for years, involve sensitive human-subjects data or proprietary partner datasets, and culminate in a thesis whose integrity must stand up to rigorous examination. The consequences of using consumer-facing AI tools without guardrails are magnified.

Consider three fault lines. First, data sensitivity: a PhD student pasting interview transcripts or patient records into a free ChatGPT session could inadvertently expose protected data. Even enterprise-tier tools often rely on contractual promises rather than technical guarantees that prompts won’t be used for model training. Second, reproducibility and provenance: AI-assisted literature reviews, code generation, or drafting can muddy the audit trail unless every prompt, model version, and post-processing step is logged. Third, supervision and examination: supervisors and examiners need clear frameworks for disclosure, attribution, and acceptable use—areas where the simplistic rules written for undergraduates collapse.

AI.RDN+ explicitly targets these gaps. Rather than starting from a blank-slate ban, it assumes that public generative AI tools are already in daily use and aims to make that use transparent, safe, and audit-ready.

The deliverable blueprint: beyond policy papers

The network’s four-year programme combines large-scale consultation with mixed-methods research across the doctoral ecosystem. The deliverables are unusually practical:

  • A living resource base cataloguing publicly available AI tools, with recommended use cases and risk mitigations for doctoral workflows.
  • Large-scale surveys and qualitative research with doctoral researchers, supervisors, examiners, and research-support staff to map uptake, attitudes, and practices.
  • Co-created training modules targeted to distinct roles—PhD researchers, supervisors, research-IT staff, and professional services.
  • A public AI.RDN+ portal publishing guidance, case studies, procurement templates, Data Protection Impact Assessment (DPIA) checklists, and microlearning assets.

These outputs speak directly to the operational needs of university legal, IT, and doctoral-school teams. Procurement and contract language, DPIA templates, logging recommendations, and short training modules are the kinds of tools that can shift institutional practice within months—not years.

Strengths and the promises that must be kept

AI.RDN+ enters a landscape already dotted with isolated university AI policies and private vendor assurances. Its most notable strength is its laser focus on the doctoral lifecycle, a space that has been underserved by sector guidance. By co-creating resources with stakeholders and embedding pilots across two regional clusters, the project avoids the trap of top-down, one-size-fits-all rules that get ignored.

The explicit commitment to practical, operational outputs—contract language, DPIA checklists, logging guidance, microtraining—separates this initiative from the many that produce only descriptive reports. And the integration with national bodies provides a ready route for scaling successful tools across the entire UK sector.

Yet the project will need to navigate several structural risks. Vendor assurances that enterprise Copilot or ChatGPT Edu won’t feed prompts into public training models are contractual, not cryptographic. AI.RDN+ must provide model contract clauses that include audit and deletion rights, not just repeat vendor marketing. Automated AI-detection tools are notoriously unreliable; examiner guidance must centre on transparency and reproducibility, not punitive detector flags. And guidance must remain a “living” asset—maintained beyond the four-year grant—because models, vendor terms, and regulation will shift rapidly.

Smaller institutions with limited capacity will need low-effort templates and microtraining that don’t assume dedicated AI-staff. And the carbon cost of widespread inference should feature in procurement advice.

The practical checklist: what IT teams can do now

For university IT, research-IT, and doctoral schools, the immediate priority list is clear—and AI.RDN+ is well-placed to validate and amplify these steps:

Governance and procurement
- Demand explicit contractual clauses preventing the use of institutional prompts or data for public model training, with deletion, audit, and breach-notification provisions.
- Conduct DPIAs for any AI service that could process research-participant data.

Controlled access and logging
- Prefer enterprise or campus-provisioned AI workspaces (licensed Copilot, ChatGPT Edu, or on-premise open-source stacks) over unvetted consumer accounts.
- Implement central logging that captures prompt, model, version, timestamp, and any post-processing—this preserves the provenance auditors will require.

Input hygiene
- Produce standard redaction templates and 15-minute training to strip personally identifiable information before any prompt is pasted into a chatbot.

Supervision and examination
- Treat AI outputs as research artefacts: require disclosure of tool names, prompt approaches, and human edits in methods sections or thesis acknowledgements.
- Give examiners checklists for interpreting AI-assisted materials without assuming misconduct; focus on transparency and reproducibility.

Training
- Develop role-targeted micro-modules: prompt literacy for PhD researchers, attribution and assessment guidance for supervisors, procurement and DPIA training for research-enabling staff.

Iterate
- Start with small pilots that measure time savings, error rates, and researcher confidence—say, literature triage or code assistance in computational labs—and scale only proven practices.

How AI.RDN+ fits the national picture

The broader sector is already moving from blanket bans toward managed adoption. Research England has backed multiple initiatives—such as the Next Generation Research SuperVision Project and Prosper—to embed responsible AI use in research culture. AI.RDN+ uniquely layers a doctoral-lens onto that movement, and its regional cluster approach allows it to test guidance across disciplines and institution types before national roll-out.

This networked, sector-integrated design increases the odds that outputs won’t languish on a shelf. But the acid test will be whether the project delivers truly reusable, low-friction assets that smaller institutions can adopt without dedicated AI teams.

What success looks like

For AI.RDN+ to leave a durable mark, it must move beyond descriptive reporting and deliver:

  • Reproducible evidence of doctoral AI usage, with published instruments and anonymised data so others can validate findings.
  • Legally usable procurement and DPIA templates that IT and legal teams can drop directly into vendor negotiations.
  • Role-specific training modules with evaluation frameworks showing that training measurably improves practice.
  • An actively maintained public portal updated as models, vendor terms, and regulations evolve, with a clear handover to sector bodies for post-grant maintenance.

If the network hits these marks, it could become a national resource that outlasts its funding window—exactly what doctoral research needs as generative AI becomes as ordinary as the word processor.

The immediate tasks are clear for anyone running university IT or doctoral programmes: lock down contracts, build controlled AI workspaces, log interactions, train people quickly, and center assessment on transparency—not detection. AI.RDN+ has the scale and the mandate to make those tasks easier, faster, and more consistent across the country. The next four years will show whether it turns that promise into tools that real teams can actually use.