Remember: 

Generative AI should only be used as a tool to enhance your learning, not replace it. Relying on AI to perform your research or complete your assessments risks leaving you with significant knowledge gaps, and may prevent you from developing a deep understanding of your topic. Overreliance on AI can also negatively impact your ability to think critically and solve problems; skills which are crucial to your success at university and beyond.  

Suggestion:

Rather than letting AI do your work for you, think of it as a tool to help you get started or to polish off your final product. For example, you could ask ChatGPT to give you some background about your topic, or Perplexity AI to find some seed references to help you begin your own independent research. Similarly, you could ask ChatGPT to suggest some ideas for a discussion, which you then develop and expand upon yourself. Doing so takes advantage of the incredible power of AI, without compromising your own learning. 

Useful Resources: 

How you should - and shouldn't use ChatGPT as a student

Remember:

Generative AI tools like ChatGPT produce outputs derived from the data they were trained on. If this data was biased, inaccurate, or incomplete, these errors or omissions may show through in a tool’s outputs. 

Some common issues that you may encounter while using AI in your studies include: 

• Inaccurate Information: 
  While generative AI tools will often produce information that sounds compelling or authoritative, they are not infallible, and have been known to present inaccurate or misleading information as fact. It's important to always approach AI generated information with a critical eye, and save yourself the embarrassment this US lawyer experienced when he presented AI fabricated cases in court as fact.
   
• Incomplete information:
  Many generative AI tools possess a knowledge cut-off date, meaning they can only draw on information up to a certain point in time. This limitation can prevent them from accessing information on recent developments or evolving topics, resulting in outputs that are incomplete or outdated. 
   
• Representational Bias:
  The training data may overrepresent certain demographics, cultures, or viewpoints, leading to outputs that lack diversity or perpetuate stereotypes. For example, an AI tool might be prone to generating more examples of male scientists or Western historical figures while overlooking less represented groups. 
   
• Historical Bias: 
  Training data often reflects past societal norms and biases, which can be reproduced in AI outputs. This may lead to outdated or discriminatory perspectives being presented as current or factual. 

Suggestion: 

While generative AI tools can be incredibly useful assistants, you must always ensure that their information outputs are correct. AI can present incorrect or misleading information in a compelling and authoritative way, so be sure to verify any generated information by comparing it to credible sources and conducting your own independent research, to ensure that it is accurate, complete, and free from bias before including it in your work.

Useful Resources:

https://www.unesco.org/en/artificial-intelligence/recommendation-ethics/cases

Remember: 

Many generative AI tools will retain your information inputs and store them for later use, often as training data, but sometimes for other purposes as well. These tools may be vulnerable to hacking or data breaches, resulting in the possible theft or misuse of your personal information. 

Similarly, AI tools may also retain the content that they generate on your behalf, potentially compromising the originality of your work or leading to unintended disclosure of your ideas. It's crucial to understand that once your personal information or ideas are input into an AI system, you may lose control over how they're used and who has access to it. 

For these reasons, it is also critical that you do not input any intellectual property that is not your own, including the intellectual property of UWA or its staff, as it is not yours to distribute. 

Suggestion:

Before inputting any of your own (or others') personal information or ideas into a generative AI tool, make sure that you understand and are comfortable with the way that data will be stored and used in the future. If you are not comfortable, or the AI tools makes it difficult to ascertain how your information will be stored and used, do not enter it.  

You should never upload any of the University's intellectual property into a generative AI tool unless you are explicitly instructed to do so. Similarly, you should never input anyone else's personal information, intellectual property, or ideas into an AI tool without their consent to do so.

Remember:

Many AI models operate as ‘black boxes,’ making it difficult to find or understand information about their internal decision-making processes, capabilities, and limitations. There is often a lack of visibility regarding their training datasets, so it can be difficult to assess how accurate or reliable a tool’s outputs are likely to be.

Similarly, you may also notice a lack of transparency when trying to determine what AI tools do with the user information they collect.

Suggestion:

Always be sure to evaluate the outputs of generative AI tools to ensure their accuracy before including them in your work. While using AI to find sources can be incredibly useful, some popular AI tools such as ChatGPT have been known to ‘hallucinate,’ and may on occasion present realistic sounding, but entirely fictitious information as fact. Evaluating the veracity of information outputs is even more important if you are unsure about a tool's processes, capabilities, limitations, or training datasets. 

You should also be wary when entering any personal information into a generative AI tool, especially if the tool makes it difficult to determine how your information will be stored and used in the future. 

Copyright

Intellectual Property and Generative AI

As a user or developer of generative AI tools or models, you are obligated to be aware of and adhere to relevant copyright and other intellectual property legislation. Copyright and generative AI generally interact in three main ways: 

1. Training generative AI models on copyright works 
2. Uploading copyright works to generative AI tools 
3. Copyright status of works produced using generative AI 

Training generative AI models 

Communities of artists, writers, academics and musicians have raised concerns regarding Gen AI models that have been trained on their work without permission and with a lack of regard for the intellectual property and moral rights of creators.  

Uploading copyright works to generative AI tools

Copyright owners have the exclusive rights to the use of their work, including the right to dictate how and when a work can be reproduced, published and communicated. Users of generative AI tools should not upload or copy any works that they do not own the copyright of into these tools. This includes documents such as unit materials, journal articles, book chapters and reports, as well as images and videos. If you are not sure whether you can use a work in this way please consult a librarian.  

Copyright status of works produced using generative AI 

In order for a work to be protected by copyright it must have a human author who contributed independent intellectual effort to its creation. Generally speaking, outputs of generative AI are not protected by copyright and users cannot claim copyright ownership of work they have produced through the use of generative AI unless it has also had significant human intervention.  

Useful Resources

Creative Commons understanding licencing and AI  

Arts Law Centre of Australia, Artificial Intelligence (AI) and Copyright

Terri Janke and Company, The new frontier: Artificial Intelligence, copyright and Indigenous Culture 

Academic authors 'shocked' after Taylor & Francis sells access to their research to Microsoft AI  

Arts Hub, Sharp rise of AI: artists’ friend or foe?