Decision making relating to the use of generative AI within course assessments at UniSQ is made at school and course level. The Library's recommendation is that students consult their course examiners about what, if any, use of generative AI is allowed for each individual course they enrol in so that they receive accurate and focused advice.
The UniSQ Academic Integrity Unit provides advice at the university level and manages the Student Academic Integrity Mandatory Training module that all students are required to complete. Within the academic integrity mandatory training module (UniSQ student log-in required) there is a statement within section 2 relating to permissions associated with the use of generative AI in university assessments.
If you have been given explicit permission to use generative AI in any of your courses, the Library can guide you with evaluation and accurate citation of these resources.
You should take a critical approach when reviewing the responses you receive to your prompts. Remember:
There are many different generative AI tools and each has been designed to do certain tasks. It is a good idea to learn more about any AI tool you may be considering using as part of your studies by reading the "about" section on the web page.
Consider:
Many AI tools have a basic, free-to-use version, but it is becoming more common for AI tools to have fee-based versions that promise improved functions and enhanced performance.
A generative AI tool is only as good as the data it has been trained on. Many tools have been trained resources that may be inaccurate, incomplete, or biased which then affects the quality of the tool's responses.
Currently, most generative AI tools cannot conduct live searches of online sources of information. For example, currently ChatGPT3.5 only searches up to September 2021. This means that although outputs may seem credible, you must always check to be sure that the information provided is current and correct.
Similarly, generative AI tools are unable to access the journals and course resources available to you as UniSQ students so can only use generally available information to generate responses to your prompts.
OpenAI. (2023).
Generative AI tools cannot read and synthesise information like humans can. Instead, they use their training and algorithms to predict what might be a convincing answer to a prompt, regardless of whether it is accurate, and can generate fake facts are known as AI hallucinations. This can include fake citations that may lead to academic integrity issues if they are then used in an assignment.
Consider:
You can learn more about generative AI hallucinations in this article from Nature.
The response generated by an AI tool is highly dependent on the quality of the prompt it is given, and may not provide the detail or depth of analysis required for university-level study. Think about the response you get from a search.
Watch the video below to learn about the trustworthiness of AI and the bias in AI-based products and systems.
National Institute of Standards and Technology. (2021, June 23). Bias in AI [Video]. YouTube. https://youtu.be/aeBLboArW8c
Many reports describe ethical concerns around AI. This includes that generative AI tools are being trained using copyrighted materials without permission from the original creators. This raises ethical issues around intellectual property and copyright infringement.
There are also concerns about transparency and data privacy with AI tools. Many people are concerned about what these tools do with the data and information they collect on users. All online resources collect some kind of data about your visit and many AI tools require you to register your email address before you can use them.
We might like to think of machine learning as rational and scientific, but AI tools can perpetuate biases that are ingrained in the data used to train the system. You can find out more from the Indigenous Protocol and Artificial Intelligence Working Group's Position Paper.
Consider: