IEEE Referencing: Generative AI

Referencing generative artificial intelligence (AI) in learning and research

Generative artificial intelligence (AI) language models (such as ChatGPT, Copilot, Midjourney) respond to natural language text inputs and are designed to generate human-like text or image responses. 

The  updated VU Academic Integrity Policy provides a definition of Artificial Intelligence and, in Part C of the Policy lists "The use of artificial intelligence models to generate any of the above without acknowledging the use of the model." as a form of plagiarism. 

Student Guidelines for using text generating tools in your work and Staff Guidelines are provided within the accompanying Academic Integrity Guidelines.

Only use these tools if explicitly directed to do so by your VU academic as part of your learning or research experience. If in doubt, check with the relevant teaching staff member or research supervisor. Using these tools without permission may be considered an Academic Integrity breach.
When permitted to use AI in your work, it is essential that you:

  • understand the limitations of the technology and the risks of using it, for for example:
    • generative AI can produce incorrect or fabricated output
    • generative AI output can contain ethically questionable content
    • generative AI output does not consider how certain words, phrases or images can cause harm to sectors of our community, such as First Nations students and staff.
    • generative AI output contains no references, and if asked to create references they may be incorrect or fabricated
  • critically evaluate any output it produces, as you would with any other academic source

Any output from a generative AI Program/App used in your work must be clearly cited using the conventions of your referencing style.

Output from generative AI is a non-recoverable source; that is, there is no way for you to link to it, and no way for your reader to access the exact output that you are referring to. The current recommendation is that you base the reference for generative AI output on the reference format appropriate for works that cannot be recovered by your reader. For example personal communication or written correspondence. 

Generative artificial intelligence (AI) is a recent development, and at the time of writing not all referencing styles are offering clear guidance, and where advice exists it is still only interim guidance. As things change and guidance becomes clearer the advice on this guide will be updated accordingly.

Updated: 13 April 2023

IEEE referencing style example

There is no formal guidance regarding how to reference AI-generated content in IEEE referencing style. Until formal advice is available use the Unpublished Material / Private Communication format. (VU Library, March 2023)

  • As the private communication may not provide recoverable data, the private communication may be cited in-text only. 
  • If citing your private communication in-text only, provide the initials as well as the surname of the communicator, along with as exact date of communication as possible.

In-text example:

Modern research tools lead to better results (OpenAI's ChatGPT, private communication, 10 March 2023).

Reference list:

You do not need to include an entry in the reference list.