Skip to Main Content
Generative AI (GenAI) has become an important tool in academic environments, offering new possibilities for teaching, learning, and research. At our institution, Microsoft Copilot is the officially supported GenAI platform, accessible with your university credentials.
Different Uses Across Academic Levels
For undergraduate students, GenAI can be particularly helpful in developing research skills and understanding complex concepts. Use it to break down difficult topics, generate study questions, or get explanations of challenging ideas in simpler terms. It can also help with structuring essays and developing arguments, though remember that the final work must reflect your own understanding and analysis.
Postgraduate students might find GenAI valuable for literature review processes, methodology development, and research planning. The tool can help identify potential research gaps, suggest relevant sources, and assist in developing research questions. It's particularly useful for getting different perspectives on complex theoretical concepts or methodological approaches.
For established academics, GenAI can streamline various aspects of academic work. It can assist in grant writing, help structure research papers, generate publication abstracts, and even help develop teaching materials. The tool can also be useful in reviewing and providing feedback on student work, though it should not replace personal academic judgment.
Best Practices when thinking about your input into GenAI
When using GenAI in academic work, always start with a clear purpose. Rather than asking broad, open-ended questions, frame your queries with specific objectives and constraints. For instance, instead of asking "Tell me about climate change," try "Explain the key debates in climate change policy from the past five years, focusing on economic impacts."
To make the most of GenAI tools, develop a habit of iterative prompting. This means refining your questions based on initial responses, asking for clarification when needed, and being specific about the type of output you're seeking. Keep a record of successful prompts for future reference, and always maintain a critical perspective when reviewing AI-generated content.
When using GenAI tools, be mindful of privacy and confidentiality. Avoid sharing sensitive personal information, confidential research data, or private institutional information in your prompts. While Microsoft Copilot is the university's supported platform with appropriate security measures in place, it's still important to exercise caution with the information you share.
These tools are meant to enhance, not replace, your academic skills and judgment. Start with simple tasks and gradually explore more complex applications as you become more comfortable. Keep abreast of university policies regarding AI use in academic work, and always prioritize learning and academic integrity in your use of these tools.
The CLEAR Framework includes five principles:
| Concise |
Use clear, simple language. |
|---|---|
| Logical | Structured and ordered approach Establish context, or give the AI a role Ensure logical flow |
| Explicit |
Define instructions and format (eg.a Table;dot points etc.) |
| Adaptive |
Be flexible – rephrase and restructure. |
| Reflective |
Carefully evaluate AI responses. Identify areas for improvement. Use insights to further refine strategies for engagement. Encourage AI to think creatively. |
Lo, L. S. (2023). CLEARer Dialogues with AI: Unpacking Prompt Engineering for Librarians [YouTube Video]. In CHOICE Media Channel. https://www.youtube.com/watch?v=3pvmMEnJhCs
Australian Catholic University Library. (2025, January 9). AI Prompts. https://libguides.acu.edu.au/ai-prompts/clear
Here are some prompt engineering strategies from OpenAI (they own ChatGPT), which are useful for any Generative AI chatbot you might come across. These mostly focus on the specificity of your prompt, and provide ways to think about what you need from the chatbot before you ask it for something.
You might need to play around with some of these tips to understand how/why they’re working, and to figure out what you need for your situation.
Write clear instructions.
Having a really clear idea of what you need out of the GenAI will help you give clear instructions – remember that the GenAI can’t read your mind, and you need to be explicit about what you want (and sometimes, what you don’t want!). You can also ask more questions and adjust as the output is presented to you. The less guessing the GenAI bot has to do, the better answer you’ll get. Here are some things to think about when writing your instructions:
Provide a reference text
GenAI can invent answers – also known as hallucinating – especially when it can’t readily access a suitable answer. Try supplying a ‘cheat sheet’, and provide a reference text so that the model can answers with fewer fabrications. This can be a text itself, or provide it with citations it can focus on.
Split the tasks you’re asking for
Like with a lot of scenarios, breaking up a task can help the GenAI give a better output. Complex tasks have higher error rates than simple tasks (in humans, too!). As you would when you’re breaking down an assignment task for your brain, break down what you’re asking the GenAI to do – a workflow of simpler tasks, in which those outputs from earlier tasks are used in the construction of the inputs for later tasks.
Give your chatbot time to ‘think’
Just like it can take us some time to work through an answer, so GenAI models could also use some time to work out an answer. There are couple of things you can do to make sure the GenAI is ‘reasoning’ towards a more reliable answer
Use other tools if you need to
There may be much more relevant tools to use for specific processes (like a code execution engine for math and coding). Compensate for the weakness of the models by feeding it output from other tools, if appropriate.
Test as you go!
Evaluate model outputs – this can be used with gold-standard answers found elsewhere, or just knowing enough about your topic before you start that you can help you evaluate the outputs as you’re going, to make sure you’re getting accurate information or questions from the GenAI.
OpenAI. (n.d.). Prompt Engineering. Retrieved January 9, 2025, from https://platform.openai.com/docs/guides/prompt-engineering
