AI in Research Principles
The use of artificial intelligence in research is evolving rapidly. This page sets out the University's shared expectations for researchers using AI and provides a framework for considering its opportunities, limitations and potential impacts. The principles support good research practice across disciplines and throughout the research lifecycle.
Artificial intelligence is rapidly changing the research landscape, creating new opportunities for discovery, creativity and impact across disciplines. To help researchers make informed, ethical, and responsible decisions about the use of AI, the University has developed a set of principles that reflect our commitment to research excellence, integrity and innovation. The principles below provide a shared framework for the responsible use of AI in research.
They are intended to support researchers in navigating both the opportunities and the challenges presented by these technologies, while recognising that practices, tools and expectations will continue to evolve. The University is developing further guidance, training and support to help researchers use AI confidently, responsibly and effectively throughout the research lifecycle.
We aim to ensure that everyone carrying out research within the University who wishes to use AI in their research activities can do so in a way that is:
Informed
Responsible use of AI starts with understanding what these tools can and can't do and innovating responsibly from sound foundations. We recognise that researchers, particularly in fast-moving fields, may wish to explore beyond established tools. Responsible innovation means anticipating and reflecting on potential impacts, engaging with people who may be affected, and acting to shape processes to mitigate those impacts. Researchers should be able to work confidently, drawing on up-to-date knowledge of the risks and institutional support available across all disciplines.
Secure
AI use sits within the University's existing data governance obligations and structures, including GDPR. The same principles that govern how we handle sensitive research data apply to using AI. Researchers should be familiar with practical strategies for minimising risk; this matters when working with confidential or personal data, where the consequences of getting it wrong are significant. Researchers should also carefully consider the intellectual property implications of inputting content into an AI tool.
Sustainable
AI has considerable potential environmental costs. The University has commitments on carbon and environmental impact, and AI use should be aligned with these. When there is a choice between tools or approaches, researchers should consider energy and natural resource demands.
Transparent
Researchers should be open about how AI has contributed to their work. This applies throughout, from how data was processed to how findings were written up. AI tools cannot take responsibility for outputs and accountability for the accuracy and rigour of information contained in research outputs lies with researchers. Transparency is essential to make AI use in research defensible and trustworthy.
Equitable
Access to AI tools and training should be available to researchers regardless of where they sit in the institution or the budget of their school or group. Beyond access, AI use in research must reflect our EDI Equity Diversity and Inclusion commitments. That means being aware of bias in inputs as well as outputs, and ensuring that the tools we use risk producing results that disadvantage particular groups.
These principles were developed by a cross-University task and finish group, and endorsed by the University Research Committee.