Module 2: Using LLMs for your studies
What You’ll Learn
By the end of this module, you’ll be able to:
- Explain how Large Language Models (LLMs) work and describe their main strengths and limitations compared to tools like search engines.
- Create effective prompts that are specific, contextualised, and ethical, distinguishing between appropriate and inappropriate uses of LLMs in learning.
- Critically evaluate AI-generated outputs for accuracy, quality, bias, and cultural context, and apply strategies to improve reliability.
- Use LLMs responsibly in academic work, including acknowledging their use appropriately and ensuring AI supports rather than replaces their own thinking and creativity.
1. How do LLMs work?
In module 1, we introduced Large Language Models (LLMs) (like ChatGPT, Claude, Copilot, or Gemini). LLMs can generate text, images, and ideas.
Think of them as a keen assistant – they can help with brainstorming, summarising, and improving your writing. But remember:
- They don’t always know if something is true. They are trying to please you and will not tell you when they don’t know the answer, Rather, they will just make it up!
- They sometimes make mistakes or “hallucinate.”
- You need to check their answers, just like you would check a Wikipedia page.
To learn more about how LLMs are trained, watch this video that takes you through a simple example using Dr Seuss’s famous poem, Green Eggs and Ham.
2. Writing Great Prompts
👉 Start with free versions like ChatGPT — that’s usually enough.
A prompt is what you type into an AI. The better your prompt, the better the result.
💡 Tips for good prompts:
- Be specific: instead of “Explain photosynthesis,” ask “Explain photosynthesis to a Year 10 student in simple terms with an example.”
- Give context: say who it’s for (e.g., “I’m writing a speech for class”).
- Choose a format: ask for a table, bullet points, or a paragraph.
- Set the tone: “Make it fun and easy to read” or “Make it formal and academic.”
🔍 Example:
“Create three quiz questions about digital footprints for Year 9 students. Provide multiple-choice answers and explanations.”
The following provides a simple guide on what may be considered inappropriate and appropriate types of prompts for using LLM GenAI.
| Not recommended – rather use a free tool like Keenious that supports finding real research articles. LLMs’ tools still hallucinate references so it should never be relied on to create a reference or literature review for you. | Appropriate Prompts: these are examples where AI is used to support the process. It allows you to scaffold AI not a replacement for your own work. | Inappropriate Prompts where the AI does the work. |
| Brainstorming ideas | “I have to write an essay about xx. Can you give me some thought starters to help direct my writing?” | “I have to write an essay about [insert topic of study]. Can you draft me this essay?” |
| Concept Explanation | “Can you explain the concept of [insert complex concept] in simple terms?” “Can you explain the difference between [insert concept]and [insert other concept]?” “what are some of the alternative theories to xxx]? | “I have to write an essay about [insert complex concept]. Please write this for me.” |
| To generate literature/references | Not recommended – rather use a free tool like Keenious that supports finding real research articles. LLMs’ tools still hallucinate references so it should never be relied on to create you a reference or literature review. | “Generate me a list of references that I can use for my research” “What academic resources could I use to research the topic [insert topic of study]?“ |
| Methodology Suggestions | “What are some common research methods used in [insert field of study]?” | “Develop an essay on [insert field of study]?” |
| Writing Assistance | NONE – When editing your work, it is better to use a free tool like Grammarly to check what is being changed. | “Can you help me rephrase this sentence/paragraph in a more academic tone?” |
| Editing work | NONE – When editing your work it is better to use a free tool like Grammarly to check what is being changed. | “Can you check this paragraph for grammatical errors?” |
3. Things to consider when using LLMs
It is important to remember that AI is not a search engine. It can make up facts or references. Rather, use Google or your library databases for reliable sources. The following are some key limitations of AI that you need to keep in mind and mitigate when using it.
| Limitation | Description | Implication |
| Factual in accuracy | LLMs are not designed to guarantee factual correctness but rather function as a kind of auto-complete (like when texting), providing responses that are deemed logically coherent rather than strictly accurate. Therefore, anything that a tool like ChatGPT might generate is not guaranteed to be accurate or even reliable. ChatGPT is known to make up information (such as references to articles). Also, while LLMs have good general knowledge and understanding of a wide range of topics, they may not have the depth of domain-specific knowledge required for your course. | It is your responsibility to check the information that is generated. Any information that you copy from ChatGPT that is factually incorrect (such as citations) and not checked, will be considered a form of fabrication. Not only is it important to quality check the information we get from ChatGPT but also be critical about the information we are getting. Don’t outsource your learning to something that doesn’t think for itself! |
| No quality control | While your writing may look good on the surface, ChatGPT lacks the ability to reason or create novel ideas and generally only covers concepts in a superficial (and uninteresting) manner. These tools are not a substitute for your critical or creative ideas/understanding, so using LLM without adding your own ideas/checking understanding will impact your marks. | Although LMS can generate coherent and context-aware responses, these are simulated, so while it may have a good general idea of bi-cultural meanings, it doesn’t understand what this really means. Also, it struggles to understand longer-term context or maintain consistency across extended conversations. It also does not understand cultural context, and therefore, while it may have a good general idea of bi-cultural meanings, it doesn’t understand what this really means. |
| Cultural and linguistic bias | LLMs are trained on text data that may contain biases favoring specific cultural and linguistic groups, often exhibiting a Western (USA) centric perspective. | Consequently, the text generated by AI language models tends to lack diverse influences from non-Western countries, resulting in generic work that may overshadow your unique voice and perspectives in your writing. It is essential to ensure that your work reflects your individuality and is not solely reliant on what can be a bland “AI-voice”. |
| Bias in AI models | LLMs are trained on vast amounts of textual data, which may contain biases present in the source material, specifically biases relating to social stereotypes and discrimination, exclusionary norms, and multilingualism (Zhuo et al, 2023) | These biases can then inadvertently permeate your writing and therefore continue to propagate these biases. |
| Contextual Understanding | Although LMS can generate coherent and context-aware responses these are simulated, so while it may have a good general idea of bi-cultural meanings it doesn’t understand what this really means. Also, it struggles to understand longer-term context or maintain consistency across extended conversations. It also does not understand cultural context and therefore while it may have a good general idea of bi-cultural meanings it doesn’t understand what this really means. | It is important that your work reflects the bi-cultural competencies for working in NZ, LLMs cannot generate. |
Remember, it is your responsibility to use AI wisely – Don’t let AI do your whole assignment. It should help you think, not think for you.
4. How to acknowledge the use of AI in your assessment
Your institution or school will tell you how they wish you to use AI in your work. However, here are some example statements that you can use to acknowledge the use of AI in your assessment.
Basic format:
I acknowledge the use of [insert AI system(s) and link] to [specific use of generative artificial intelligence]. The prompts used include [list of prompts]. The output from these prompts was used to [explain use].
Examples
| I acknowledge the use of [1] ChatGPT (https://chat.openai.com/) to [2] generate materials for background research and self-study in the drafting of this assessment. I entered the following prompts on 4 January 2023: [3] Write a 50 word summary about the ethical implications of using Generative AI in higher education. Write it in an academic style. Add references and quotations to justify statements. [4] The output from the generative artificial intelligence was adapted and modified for the final response. |
| I acknowledge the use of [1] ChatGPT (https://chat.openai.com/) to [2] generate materials that were [4] included within my final assessment in modified form. I entered the following prompts on 4 January 2023: [3] Write a 50 word summary about using Generative AI in higher education. Write it in an academic style. Add references and quotations to justify statements. |
| I acknowledge the use of [1] ChatGPT (https://chat.openai.com/) [2] to refine the academic language and accuracy of my own work. On 4 January 2023 I submitted my entire essay (link to google document here) with the instruction to [3] “Improve the academic tone and accuracy of language, including grammatical structures, punctuation and vocabulary”. [4] The output (here) was then modified further to better represent my own tone and style of writing. |
| I acknowledge the use of [1] OpenAI Codex ( https://openai.com/blog/openai-codex /) [2] to refine my code. On 4 January 2023 I submitted my xxx |
4. AI in Your Studies – Tips
✅ Start simple – free versions (like ChatGPT) are usually enough.
✅ Write clear prompts – be specific about the task, audience, format, and tone.
✅ Use AI for support, not shortcuts – brainstorm ideas, clarify concepts, or refine writing, but do the thinking yourself.
✅ Check accuracy – always verify facts, references, and examples against reliable sources.
✅ Keep your voice – mix AI’s suggestions with your own words to avoid “AI-generic” writing.
✅ Be critical – watch out for bias, oversimplification, or missing context in AI’s responses.
✅ Stay ethical – follow your institution’s rules and acknowledge AI use where required.
✅ Protect your learning – don’t let AI replace your effort; use it to enhance your skills. Use AI to brainstorm, draft, or check ideas — but add your own thinking.
5. Key Takeaways
- LLMs are powerful assistants, not knowledge experts. They generate text by predicting patterns in language, not by understanding facts — so always double-check what they produce.
- Good prompts make a big difference. The clearer and more specific your instructions, the more useful and relevant the AI’s responses will be.
- Use AI to support, not replace, your learning. Brainstorming, simplifying concepts, or refining writing are appropriate uses — but outsourcing whole assignments to AI is not.
- AI has limitations and biases. It can “hallucinate” information, reproduce stereotypes, and lack cultural nuance, so critical evaluation is essential.
- You are responsible for how you use AI. Blend AI’s input with your own ideas, and acknowledge its use properly in assessments.