Artificial Intelligence and academic integrity

Artificial Intelligence tools such as ‘ChatGPT’ are becoming pervasive. What are the learning and teaching implications and how is ACU responding?

What do Artificial Intelligence language tools do?

‘ChatGPT’ and similar tools are language models that produce output in response to prompts. ChatGPT is fundamentally a more sophisticated version of the predictive text function on a smartphone. It cannot think or analyse – it merely synthesises a response using algorithms running on word sequences from its (admittedly large) databanks.

ChatGPT generates text in response to specific prompts that a user enters and can answer questions on a variety of topics. It can critique text to a limited extent (eg, check reference formatting), and it is useful for generating and debugging code.

However, it is not as efficient at evaluating sources or comparing perspectives. As it is essentially a predictive text tool, the output is phrased confidently, but it can provide incorrect information, and will generate false information to complete an output (eg, fabricated but superficially plausible references).

How might these tools be used?

Artificial Intelligence (AI) tools are here to stay. Undoubtedly, over the coming months to years, their use will be embedded in our practice. We might use such tools to generate a first draft of a cover letter or job description, or as a starting point summary when researching a new field.

If their use becomes embedded in the professional practice of our graduates’ workplaces, we need to ensure that our students are trained in how to use these tools effectively and ethically, and how to acknowledge their use where required.

Unfortunately, some students may attempt to generate outputs that they then inappropriately submit for assessment as their own work.

The actions ACU is taking

Under proposed amendments to ACU’s Academic Integrity and Misconduct Policy, unauthorised use of artificial intelligence tools, and other tools such as paraphrasing and translation tools, would be considered to be a type of academic misconduct. Utilising these tools to produce assessed work would be a form of contract cheating as it involves engaging a service to prepare part or all of an assessment task.

Students must not use these tools to prepare assessable work unless their use is specifically authorised. If in doubt, they should check their Unit Outline and seek clarification from their Lecturer-in-Charge (LIC).

Individual discipline areas may decide to allow their use in a particular unit or under specific circumstances. This will need to be clearly communicated to students. Any authorised use of the tools will need to consider aspects such as:

• Academic integrity: How do students acknowledge their use of AI tools? How do they indicate which parts of their work have been derived from an AI tool?
• Equity: Is access to the tool available and possible for all students?
• Pedagogy: Does the tool support individual and collaborative learning and enable students to set enhanced goals for achievement?

How do I improve my assessments?

ACU is providing staff with workshops to discuss how to alter assessment design to maximise security of assessments in the face of potential illegitimate use of these tools. The first workshop will run in Week 2 and 3 of Semester 1.

You can register through one of the links below.

What can I do in my teaching to ensure that the availability of AI apps does not compromise academic integrity?

Session A: Friday 10 March, 2pm–3pm (AEDT)
Session B: Wednesday 15 March, 11am–12pm (AEDT)

On completion of this workshop, teaching staff will have selected strategies for working with students and assessments in their units given the current availability of ChatGPT (and other AI apps).

Do you have a particular interest in Artificial Intelligence tools?

ACU is establishing a Community of Practice to encourage discussion around emerging practice, including acceptable use, ethical implications, and approaches to assessment. If you are interested in becoming involved, please join the AI Community of Practice Teams site.