Advice to students about using generative Artificial Intelligence

The Academic Integrity Policy for Coursework Awards stipulates that Academic Misconduct encompasses the use of artificial intelligence tools (including, but not limited to, ChatGPT, Microsoft Copilot or similar), in any assessment task, including project or thesis.

Section 5.1(g) of the Policy notes there are three exceptions and a general requirement:

Exception 1: to conduct preliminary research on a topic. Preliminary research is a step prior to writing. Acceptable AI use for preliminary research includes use for an initial topic search, to clarify key terms, to identify useful starting readings, translating readings into another language, or to summarise a reading to ascertain its relevance. Non-acceptable AI use for preliminary research use includes using a tool to provide answers to the assessment task, or a paraphrase of the assessment task, and using any generated content from AI directly in a submitted piece of assessment. Where used for preliminary research the name and version number or date of the tool(s) used (e.g., “ChatGPT Version 3.5”) must be declared in text or in an Appendix;

Exception 2: to obtain general advice on spelling and grammar. However, students should not use explicit AI features added into a grammar tool; for example, using Word or Grammarly for grammar advice is acceptable, but using an additional AI tool added into a grammar tool, like GrammarlyGO, is not acceptable). Students writing in a language other than their first language can use AI for general advice on spelling and grammar, but not for translating their written work;

Exception 3: in the case where a lecturer provides written authorisation for the use of an artificial intelligence tool for defined aspects of an assessment task, in which case any student use must be declared and fully referenced. The name and version number of the tool(s) used (e.g., “ChatGPT Version 3.5”) must be declared and an Appendix showing student prompts to the AI and the AI responses must be included. Written authorisation can be provided in assignment instructions.

General Requirement

All students must use the assessment coversheet required by their Affiliated College, and keep drafts of all assessment tasks in order to provide these to lecturers if concerns are raised about potential misuse of AI tools (NB: software such as Word automatically keeps a version history containing drafts). NB: The University uses software (Turnitin) to detect potential cases of plagiarism and misuse of AI tools.

NB: AUT Affiliated Colleges are free to stipulate further details about AI use – students should contact their college for advice. 

Adapted from Mark Kulikovsky, BCSA

AUT students are to acknowledge any use of AI by adding a Declaration such as the following to their Assessment’s title page:

Example for Exception (1):

I acknowledge the use of ChatGPT Version 3.5, https://openai.com/chatgpt to generate materials for preliminary research for this assessment task.

I affirm no content generated by any AI technology/ies has been presented as my own work. When ChatGPT [or insert other AI] is used as a resource it is duly referenced.

Example for Exception (2):

I acknowledge the use of ChatGPT Version 3.5, https://openai.com/chatgpt to obtain general advice on spelling and grammar for this assessment task.

I affirm no content generated by any AI technology/ies has been presented as my own work. When ChatGPT [or insert other AI] is used as a resource it is duly referenced.

Example for Exception (3):

I acknowledge the use of ChatGPT Version 3.5, https://openai.com/chatgpt to assist with this assessment task in accordance with the written authorisation of my lecturer.

I affirm no content generated by any AI technology/ies has been presented as my own work. When ChatGPT [or insert other AI] is used as a resource it is duly referenced.

Additionally, students are to include the details of ChatGPT material in an Appendix, for example:

The following prompts were used in ChatGPT:
Prompt 1: What were the key elements of the Arian Controversy?
ChatGPT Output 1: [insert ChatGPT text]
Prompt 2: What were the theological implications of the Arian Controversy?
ChatGPT Output 2: [insert ChatGPT text]
The outputs from ChatGPT were used, or modified, in my essay in the following way/s:
[Describe how it was used, for example: The information was used to …; multiple outputs were generated and the results merged into …; etc.]

• APA
• Chicago

Yes. The Turnitin system integrated in Moodle that is used to check for plagiarism also detects the use of AI, and inappropriate use of AI that is detected can be treated as a case of plagiarism under AUT rules (see below). The Turnitin system has a high rate of detecting essays that use AI (even just a few sentences), and it rarely makes an incorrect conclusion (Turnitin states that only 1 in 100 essays that are fully human written are incorrectly designated as AI written where over 20% of the text appears AI written). Turnitin’s AI detection score is only shown to lecturers (not students).

Any unauthorised or unacknowledged use of AI in any assessment is a form of cheating and, as such, is considered to be wilful academic misconduct. The University treats any case of suspected cheating seriously. Where unauthorised use is suspected, principles of procedural fairness and the requirements of the Academic Integrity Policy for Coursework Awards will be applied (or Research Integrity Policy for HDR).

A student submits an assignment. Turnitin reports 24% as AI generated, and identifies the relevant sentences. The staff member reports a case of suspected unauthorised or unacknowledged use of AI to the Academic Dean. The Academic Dean reviews the evidence and follows the procedure below:

1) The Academic Dean writes to the student and requests to be provided with the draft/s or notes of an early version of the assignment for review

2) If required, the Academic Dean meets with the student in order for them to explain their understanding of the assignment’s requirements, how they prepared the assignment, such as how they developed their argument/s, what sources they used, how they reached their conclusion/s.

3) The Academic Dean makes a determination on the case and informs the student in writing

4) If unauthorised use is proven, the provisions of the relevant Policy can apply.

It is important to recognise the limitations of AI, for example, ChatGPT’s GPT-3.5 data is limited to 2021 and before.  More recent versions may have access to more recent data (e.g., GPT-4o is trained on data up to June 2024, while certain enterprise AI versions have access to some real-time data).

Macquarie University has identified some of the important Limitations of AI

Accuracy – the output of generative AI tools may be inaccurate or contain false or invented information.

Privacy – in general the input and output of the tool is recorded by the organisation offering the tool. The information may become public or shared with other organisations.

Intellectual Property – be sure to read the terms of service for each AI tool to check the licence terms that cover input and output of each tool you use.

See also advice from

(a) TEQSA: Artificial Intelligence: Advice for students

(b) Turnitin: Ethical AI use checklist for students