Generative AI guidance
What are Generative AI tools?
Generative AI tools are a type of software that automatically generate content based on questions or prompts input by the user. They include ChatGPT, Bard and Synthesia, among many others. Tools can generate text, code, images, and other types of content. Generative AI tools that deal with text are also called Large Language Models, or LLMs and continuously build a ‘knowledge base’ of information by collecting users’ previous questions or prompts as its base of data. Generative AI tools can be appealing to use in that they can provide or create content quickly but have limitations for use in any academic work in that that the outputs may not be wholly reliable.
Acknowledging and referencing and generative AI tools
You should include a statement to acknowledge your use of generative AI tools for all assessed work, in accordance with guidelines from your department or course team.
This statement should be written in complete sentences and include the following information:
- Name and version of the generative AI tool used; e.g. ChatGPT-3.5
- Publisher (name of company that provides the AI system); e.g. OpenAI
- URL of the AI tool
- Brief description (single sentence) of the way in which the tool was used
- Confirmation the work is your own
I acknowledge the use of ChatGPT 3.5 (OpenAI, https://chat.openai.com/) to generate an outline for background study. I confirm that no content generated by AI has been presented as my own work.
Further requirements may be stipulated for a particular piece of assessed work and must be made clear to students when it is set. Additional requirements may include expanded description in the ‘Acknowledgements’ or ‘Methods’ section, such as:
- If relevant, the prompt(s) used to generate a response in the AI system.
- The date the output was generated.
- The output obtained (e.g. a ‘link to chat’ if ChatGPT, or a compilation of all output generated as an appendix).
- How the output was changed for use or incorporation into a piece of work (e.g. a tracked-changes document or a descriptive paragraph).
Referencing guidance is provided for Harvard and Vancouver referencing styles on the following pages:
Considerations for use of generative AI tools
There is no agreed AI Literacy framework yet in the Higher Education sector to outline necessary skills and competencies for students and academics when using these tools. Library Services has compiled a list of considerations below for the use of generative AI tools.
These considerations are organised according to the continuum of information use and management the library supports. They are not intended to be exhaustive. For further information consult your course team or tutor and your Subject Librarian. The College Academic Misconduct Policy and Procedure addresses the use of generative AI tools and academic integrity should be understood to encompass appropriate use of these tools.
- It is acceptable to use a generative AI tool as you would a search engine, for example, to look up background information. Any information later used must be referenced and cited properly. Please refer to the Library’s reference management pages.
- Generative AI tools can provide a starting point for research. We do not recommend using generative AI for in-depth research or instead of using credible sources such as academic databases. To identify and use the best academic sources for your work, your Subject Librarian can provide guidance and advice.
Critically evaluating information
- Generative AI tools’ outputs are, fundamentally, not reliable nor trustworthy as pure statements of fact. AI ‘hallucinations’ occur when factually incorrect information is included in the generative AI tool’s outputs. Therefore you cannot trust the tool as the sole source of any facts and it is essential to verify any statement generated by these tools.
- Your subject librarian and the library provide in-depth content and advice and guidance on critical evaluation of sources and the best sources to use for academic work.
- Using generative AI tools to analyse numeric data is not recommended at this stage.
Ethically using information
- It is important to familiarise yourself with your department’s current guidance on the use of generative AI tools in support of academic work. As stated in College guidance, your department may choose to invite a random selection of students for ‘authenticity interviews’ on submitted assessments.
- You may be asked as part of your assignment or assessment to use generative AI tools. We recommend that clarity is sought from your course team or tutor if you are unsure how to do this. Where there is no explicit instruction to use generative AI tools, it would not be considered acceptable to use them to write your assessed work.
- It is important to disclose the use of generative AI tools for academic work according to your department’s requirements. You can also refer to the library’s guidance on proper citing and referencing on the reference management pages.
- Unless explicitly authorised to use as part of an assessment, the use of generative AI tools to create assessed work can be considered a form of contract cheating, which is addressed on the College Plagiarism, Academic Integrity & Exam Offences web page as well as within the library’s Plagiarism Awareness courses.
- It is not advisable to add sensitive data (such as student names or other personal data) into generative AI tools, such as ChatGPT, as queries are stored and become part of the training data it draws upon.
Creating and communicating information
- When considering the use of generative AI tools for creating original academic work, ask yourself if it would be acceptable to ask another person to do this work for you. If the answer is no, then it is not acceptable to use these tools.
- Consider that outputs from generative AI tool outputs can be quite generic. You risk losing your own original voice when using it to create written work. Using these tools uncritically would likely result in a text that adopts an academic style superficially but contains little in the way of substantial communication.
- It may be unacceptable to use an essay outline produced by generative AI tools.