Guidance on the responsible use of generative AI in postgraduate research (PGR)
Principles for the responsible use of generative AI in postgraduate research
As a postgraduate researcher:
You are accountable for exercising critical judgment and oversight when using AI.
You need to be clear on why and how you are using AI in your studies and your research.
It is mandatory to be transparent about your use of AI.
Unless you acknowledge the use of generative AI or any contribution other than your own, it will be assumed that the research you present and work you submit for your programme is entirely your own.
Referring to the University’s expectation for responsible research should also inform your use of AI.
Your research must adhere to ethical standards, data management (including data protection) and intellectual property rights (e.g. copyright and ownership). Research should be authentic, avoiding the fabrication or falsification of your results. Research Integrity and our Code of Practice[1] should inform your use of AI, ensuring your work is your own, with all sources acknowledged.
Using AI in your research should align with your researcher development plan, the research culture of your wider academic community, and the Bangor University Research Development Concordat.
Undertaking a PGR programme involves more than conducting your research. It also means gaining a broader understanding of your chosen field, refining your critical thinking skills and engaging with members of the research community. Understanding the values and principles at scale, and deploying AI responsibly in line with these values, is important.
The limitations of using generative AI
- AI can ignore intellectual property (IP) rights and cause a breach of IP rights
- AI outputs can be false – outputs from the models can be wrong or nonsensical
- AI data can be fabricated – AI can produce data or ‘hallucinations’ which is information or evidence that is essentially made up on the AI platform
- AI is prone to bias
- AI is not a comprehensive source of information and may miss key information
- AI is a tool and is therefore affected by the capabilities and competencies of the user
The risks of using generative AI
Using generative AI in research can cause cultural divides within the research community, with many academics voicing concerns about the erosion of critical thinking, cyber security risks, and impacts on environmental sustainability – with AI placing increasing energy demands globally. Many generative AI systems like Large Language Models (LLM) (for example, ChatGPT) are open source, meaning the confidentiality of data entered cannot be guaranteed.
As a postgraduate researcher, it is your responsibility to maintain a high standard of academic and research integrity and to adhere to Bangor University’s principles for the responsible use of generative AI in postgraduate research (PGR). The use or misuse of AI in your PGR could lead to you committing academic or research misconduct .
Table 1: Risks of using AI in PGR, potential impacts and mitigations
Risk | Impact | Mitigation |
| Plagiarism, False Authorship & Misrepresentation: presenting AI work as your own. | Disciplinary action, including potential failure of assessments or expulsion. | Clearly acknowledge AI and follow the guidelines for the responsible use of AI in research. |
| Inaccuracy & Bias: AI models can produce entirely fabricated (‘hallucinations’) misleading, incorrect or biased outputs. | Lead to poorly evidenced hypotheses and conclusions, undermining academic credibility. Bias in the models can reinforce stereotypes. | Only use AI as a tool and not as a definitive source of information, or synthesis of information. Apply critical evaluation and check sources are trusted. |
| Breach of Duty of Care: uploading sensitive or unpublished data into an AI model may violate data protection and contravene ethical standards or national legislation (e.g. Export Control Act). | Potential data or legislative breaches, ethical violation, loss of funding and reputation. In extreme cases (National Security Investment and Export Control Acts) it could lead to prosecution & imprisonment. | Do not upload unpublished or sensitive data into open-source AI tools, unless they have been approved by Bangor University’s Digital Services and are secure. |
| Lack of Transparency & Explainability: it is not always clear where AI data has come from | This limits your ability to adhere to the principles of open research and reproducibility | Always document your use of AI tools and complement the process with your own critical analysis. |
| Overreliance of AI Tools: relying on AI to generate ideas synthesise and analyse data can limit the development of your own critical thinking skills | Leads to a lack of innovation, originality, rigour and independent researcher development | Use AI as a supportive tool with adequate checks and balances ensuring you still engage deeply with your subject and research |
Using Generative AI wisely
- Familiarise yourself with AI tools and what the University offers in terms of generative AI
- Involve your PGR supervisor in your work and intention to use AI.
- Consider if software packages you are using have integrated AI into the software (e.g. Copilot is integrated into Microsoft products).
- Read the University’s relevant policies, for example, Research Integrity; Guidance for the Responsible Use of AI in research: publishing and reviewing; and Trusted Research.
- For formally assessed work, refer to the Artificial Software and Academic Integrity: a short guide for students here before using generative AI in the production or delivery of formally assessed work.
- Evaluate all AI outputs. Always apply critical thinking and assessment when using AI outputs in your work – it is a tool. You are responsible for using any AI output, and it is not enough to cite the AI model as a definitive resource like you might a peer-reviewed publication.
- Consider the ethical approvals that could apply when integrating AI in your research methods and design.
- Ensure you have a sound data management plan and are aware of any risks of using AI in terms of data confidentiality.
- Always keep records of AI use and acknowledge where and how it was used in your work.
More guidance and how to reference the use of Generative AI for PGRs[2]
- Develop a prompt / response repository
- How to reference generative AI outputs
- How to reference the use of generative AI as a tool
- How to reference the use of generative AI as a proof-reading tool
- How to reference other uses of generative AI use