AI tools can be powerful collaborators in the academic writing process. They can assist with tasks such as:
Various tools are available for writing tasks, including ChatGPT, Perplexity AI, Microsoft Copilot, Google Gemini, and RMIT-approved VAL.
VAL is a free text-based generative AI tool available to RMIT staff and students. It ensures that the data you provide remains private, confidential, and secure. Unlike some other tools, VAL does not share your data with OpenAI or any external organisations. This safeguards your information and intellectual property, reducing the risks of misuse or misconduct.
By seeing AI as a virtual assistant, you can leverage its capabilities to enhance your research writing. RMIT RePAIR (Research Practice with AI at RMIT) provided valuable advice on using Generative AI in four general stages of research writing, from planning a paper or thesis to submitting for review or examination.
Engaging thoughtfully with AI requires careful consideration of its role and impact on your academic work, ensuring that you maintain academic integrity and critical thinking. It is important to use AI as a tool to support your efforts rather than replace your own understanding and analysis, allowing you to remain in control of the writing process.
Drawing on RePAIR's foundational insights and principles, this library guide offers practical strategies for making effective use of AI in research writing. Like any collaboration, the results are highly dependent on the clarity and detail you provide as well as how you interact with AI and refine the outputs.
Before using any AI tool, it is crucial to have a clear understanding of the result you are aiming for. Whether you are looking for a summary, a list of ideas, or help with improving a section of your thesis, knowing your desired outcome will help you formulate focused instructions and guide your further interactions with AI.
Some questions that can help you understand your needs include:
This preparation phase ensures that the AI tool’s output is aligned with your objectives and enhances the overall quality of your research work.
Once you know the output you want, you will need to guide the AI tool, similarly to how you would guide a human research assistant, by providing prompts. AI tools can handle various types of prompts to generate different outputs:
This guide will focus on text-based prompts, as they are most relevant to research writing. Writing effective text-based prompts (sometimes called prompt engineering) is a skill that requires practice and strategies to achieve the best results. A well-structured prompt can transform a vague response into a relevant and useful one. However, your prompts do not have to be perfect from the start. Feel free to experiment and adjust! If the initial response is not quite right, refine your prompt accordingly.
AI-generated content often requires a human touch to reach its full potential. While AI can generate text based on patterns and data, it does not “understand” the content in the way humans do. This is where your judgment and critical thinking come into play. It is essential to evaluate the accuracy and relevance of all AI-generated content. Use follow-up prompts to clarify, refine, or expand on specific points until you get the desired output.
The iterative process of interacting with AI is key. AI can be a valuable assistant, but it is not infallible. Therefore, it is important to continue the “conversation” with the tool to guide it toward producing content that fits your objectives and standards.
Keep in mind that once you submit your work, you are responsible for its correctness and appropriateness, including its AI-generated content.
When interacting with AI, it's crucial to provide clear and detailed instructions to ensure the AI delivers outputs that meet your specific needs. Below are important components of effective prompts:
Define the role: Clearly articulate the role you want the AI to assume. This helps the AI focus on the specific expertise or style you require, leading to more accurate and useful outputs.
For example: You are a skilled editor tasked with enhancing the readability and quality of academic papers.
Provide detailed context: Offering comprehensive context helps the AI understand your requirements better. Instead of a single sentence, use paragraph(s) to elaborate on the background and/or your objective.
For example: I am writing a research paper on the impact of climate change on coastal ecosystems, specifically focusing on how rising sea levels and increasing temperatures affect biodiversity in these areas. The paper aims to explore both the short-term and long-term effects on various species and habitats. Could you please suggest an outline for the literature review section?
Make a request: This component is often the core of the prompt, directly assigning a task to the AI. Prompts that lack a request generally serve as fillers, such as "Great, thank you!", or provide additional information for a previous request, such as "By 'schooling', I mean formal education from Foundation (aka. kindergarten, pre-school, or preparatory) to Year 12". The request component usually appears as a question or command, but sometimes implicit requests.
For example:
[Question] What are the challenges faced by international students in Australian high schools?
[Command] Suggest an outline for the literature review section of a journal article focusing on researchers’ perceptions of research integrity.
[Implicit request] I am looking for some effective ways to work from home during children’s school holidays.
Implicit requests might not always yield the desired response as AI algorithms may not accurately predict nuances of language and context.
Define the output: Clearly specify what the output should look like, including tone, length, format, and style.
For example: Summarise this article in 300 words, using bullet points. Ensure the summary is concise and informative, maintaining a professional and neutral tone. Use clear and straightforward language, avoiding jargon. Highlight key points and main arguments, ensuring each bullet point is a complete thought and contributes to a comprehensive understanding of the article's content.
Provide examples: While optional, providing examples can guide the tone or structure of the output. You might share a well-written excerpt to illustrate the tone you want to replicate or provide a sample structure to ensure the output follows the desired format. This can be particularly useful when you're aiming for a specific academic style or format. However, be cautious when providing examples, especially regarding copyrighted material, confidential information, and your own intellectual materials.
For example: Below is an excerpt of a well-written summary. Please replicate this tone and format for the given content.
Including all these elements is not always necessary, but clear instructions, detailed context, and well-defined expectations can enhance the effectiveness of AI interactions. Crafting effective prompts often involves testing and refining, as the success of various techniques may vary depending on the task and the AI tool used.
To switch to a completely different topic, click on "New Chat" or use a prompt like "Ignore everything above…" if you prefer not to draw on any previous conversation.
Break down a complex request into multiple prompts to ensure each part of the task is addressed more thoroughly and accurately. For example:
RMIT's Artificial Intelligence - Referencing Guidelines offer detailed instructions on acknowledging AI tools in your work and referencing AI-generated content. The key principle is to accurately reference any ideas or outputs generated by AI to uphold academic integrity. For HDR candidates, remember that when submitting your thesis, you will need to include a prescribed declaration statement, part of which reads:
I certify that except where due acknowledgement has been made, this research is that of the author alone; the content of this research submission is the result of work which has been carried out since the official commencement date of the approved research program; any editorial work, paid or unpaid, carried out by a third party is acknowledged; and, ethics procedures and guidelines have been followed. [emphasis added]
"Submitting for examination" by RMIT University
To ensure accurate acknowledgment of AI use, follow these best practices:
For example: On 20th January 2025, I used the pro version of GPT-4 to help draft an outline for the literature review on the use of solar energy systems in Australia. The prompt used was: "Create an outline for a literature review on the adoption and challenges of solar energy systems in Australia, including policy, technological advancements, and market trends".
For more information on referencing AI-generated content according to the various referencing styles used at RMIT, please refer to RMIT's Artificial Intelligence - Referencing Guidelines.
RMIT University (2025). Submitting for examination. https://www.rmit.edu.au/students/my-course/research-students/submission-examination/submitting-your-thesis
Effectively leveraging AI tools requires more than simply generating responses; it involves a thoughtful approach to refining and enhancing these outputs. While AI can significantly aid your research and writing processes, it is crucial to view its outputs as starting points. You need to conduct additional research beyond merely asking questions to an AI tool.
To optimise AI-generated outputs, consider the following strategies, taking into account the specifics of each task:
Review and edit: Critically evaluate the AI's response. Revise it to align with your voice and the academic standards of your discipline. Ensure the language matches the intended tone, whether formal or informal, and reflects your personal writing style. Clarify any ambiguous or vague statements to make the content precise and easily understood by the intended audience.
For example: The language is too informal. Please use a more formal and professional tone appropriate for an academic conference in the STEM field.
Expand on ideas: This is an important step in refining AI-generated content, particularly when certain sections may be underdeveloped or lack sufficient detail. Begin by identifying areas in the AI's response that seem superficial or incomplete. Look for sections where additional information or elaboration is needed to fully convey the intended message or argument.
For example: Please expand on the second bullet point, discussing and providing examples of potential challenges.
Request explanations: Ask the AI to explain its reasoning if you're unsure about the output. By seeking explanations, you can better assess the validity and relevance of the information, identify any potential gaps or errors, and make more informed decisions about whether or how to use the AI's output.
For example: I received the following output, but I'm unsure about how the conclusion was reached. Could you explain the reasoning or steps you took to arrive at this result? Please include any assumptions or key pieces of information that were considered in the process. The output I want you to explain is...
Fact-check: AI can generate information that sounds plausible but may not be accurate. Cross-checking facts against reliable sources is essential to ensure the accuracy and credibility of the information provided. This practice helps prevent the dissemination of misinformation and supports informed decision-making based on verified data.
Add human elements: Inject nuance, tone, and personal insight that AI tools may miss. For instance, when preparing a conference presentation with AI assistance, you could include a personal experience from a lab experiment or fieldwork to illustrate a scientific concept or viewpoint. This adds depth and authenticity to your work, making it more engaging and relatable to your audience.
In conclusion, while AI tools can be valuable in generating content and assisting with research, it is essential to apply a critical and reflective approach to their outputs. Remember, AI is a helpful tool, but the human element - your expertise, judgment, and creativity - remains crucial in producing high-quality, meaningful academic work.
This Library guide by RMIT University Library is licensed under a CC BY-NC 4.0 licence, except where otherwise noted. All reasonable efforts have been made to clearly label material where the copyright is owned by a third party and ensure that the copyright owner has consented to this material being presented in this library guide. The RMIT University logo is ‘all rights reserved’.