Artificial intelligence (AI) is a rapidly-evolving field, influencing many aspects of academic work, including how we manage and organise references. While AI tools are beginning to integrate with reference management software, the extent and effectiveness of these integrations vary. Please note that this guide is meant to act as a starting point and is subject to change.
On this page, you'll find information on what AI is, the University's polices regarding AI-generated content, and further signposting towards a recommended source on how to cite AI.
AI refers to computer systems that can learn from data, identify patterns, and make predictions or perform tasks that historically only a human could do. AI systems are trained on large amounts of information, such as online text and images, which is why generative AI (GenAI) tools such as ChatGPT can create new content that appears "human-like".
These tools, however, may generate inaccurate content, as well as reproducing the biases contained in the data they were trained on.
The University of Oxford maintains that AI can be a supportive tool in learning, so long as it’s used ethically and appropriately. However, AI can be used within assessments only where prior authorisation has been given, or when AI has been agreed as a reasonable adjustment for a student’s disability.
Whenever AI is used, similar safeguards to those on plagiarism should be adopted, with authors never passing off ideas or text from AI as their own, and a clear acknowledgement in an author's work of how AI has been used.
College, departments, and tutors may have their own policies on the use of AI, and it may be forbidden in local guidance. Consult with relevant staff members before using AI in your work.
Some, but not all, referencing styles now have guidance on how to format AI references. Any guidance is interim advice only, and likely to be modified in the future.
Cite Them Right is an online referencing platform that's free to members of the University. It offers advice on how to reference GenAI correctly across multiple styles, including Chicago, Harvard, and Vancouver.
In the absence of guidance for your specific AI tool, adapt the referencing style for the source type closest to the AI tool or AI output e.g. audio recording, image, personal communication. Additionally, consult tutors or relevant staff members, especially if they require you to provide a copy of the AI work as an appendix to your own.
Currently, reference managers do not have a specific resource type for AI. Again, it’s necessary to match the closest pre-existing resource type (e.g. artwork, dataset, video recording) when creating citations.
This approach ensures consistency while we wait for these tools to potentially update their resource categories to include AI-specific options.