Part of what a university education teaches is certain academic skills, such as assimilating information, constructing an evidence-based argument and expressing your thoughts in clear, coherent prose.
AI tools cannot replace human critical thinking or the development of scholarly evidence-based arguments and subject knowledge that forms the basis of your university education.
You can make use of generative AI tools (e.g. ChatGPT, Claude, Bing Chat and Google Bard) in developing your academic skills to support your studies. Your ongoing critical appraisal of outputs by reviewing them for accuracy will maximise the potential for AI outputs to be a useful additional tool to support you in your studies.
In some instances academic staff, departments and colleges may give more detailed guidance on how they expect AI tools to be used (or not used) for different tasks or on specific assignments. You should always follow the guidance of your tutors, supervisors and department or faculty.
Ethical use of generative AI tools
Your ethical use of AI tools to support your learning is paramount to ensure you uphold the standards of academic rigour and academic integrity expected of you as a student studying at Oxford.
Students using AI during their studies must learn and practise the same academic skills of note-taking and clear attribution which are safeguards against plagiarism, ensuring clear differentiation of their own work from any text or material derived from any AI tools. Unauthorised use of AI falls under the plagiarism regulations and would be subject to academic penalties in summative assessments.
Where the use of generative AI in preparing work for examination has been authorised by the department, faculty or programme, you should give clear acknowledgment of how it has been used in your work.
The advice in this guide is in line with relevant university policies, which have been updated to provide examples of unauthorised uses of generative AI tools. Guidance on plagiarism is available from the Study skills and training webpages. The policy on the Use of Third Party Proof-readers is also relevant.
Five things to think about when using generative AI tools
- How can generative AI tools be useful in supporting your learning?
- How can generative AI tools be useful in developing your academic skills?
- How (and when) can you draw on AI outputs (knowing that different generative AI tools provide different outputs and that the same tool with the same prompts can produce different outputs)?
- How can you manage the risk of false information and fabrication?
- How can you ensure you maintain good academic practice?
Six tips to keep in mind when using generative AI tools
- Always cross-check AI generated outputs against established sources to verify accuracy and identify erroneous information.
- Give significant contextual information when asking questions or prompts and ask several follow-up questions to refine responses.
- Use personae in your prompts e.g. “I am an undergraduate student who is revising for a first-year calculus exam”.
- Give examples of the kind of responses you want.
- AI tools are not good at calculations so use other established tools, calculators, Excel or Mathematica.
- Do not share sensitive personal data such as financial details or passwords with AI tools. Avoid sharing your own or others intellectual property such as patents, trademarks, designs, sensitive information, or content created by others into any AI tools.
Five ideas for academic reading
Generative AI tools may be useful in supporting you to develop your academic reading skills. However, generative AI may, in some cases, undermine development of your academic reading skills (e.g., asking an AI tool to summarise an article rather than undertaking the task yourself).
Here are five ways you can use AI to support your academic reading, suggestions for how you might provide useful prompts, what some of the limitations are of the AI generated outputs, and how to engage critically with them to augment your learning.
- When reading a paper ask for a table of key terms or outline key points in the paper. Do this yourself before asking AI to do this and compare your terms or points. The AI tool can help you build a cognitive scaffolding of your reading of a paper but you cannot rely on it, so ensure you read the paper yourself.
- Ask AI to generate thought-provoking questions based on article content. You can develop your own understanding of an article by answering the questions asked. You could also use the questions to develop your own questions in relation to the article to deepen your learning.
- Ask AI to translate sections into another language. You can ask AI to translate text into your own language. Be sure to check the accuracy of the translation and that no significant meaning has been lost in translation. This could support your understanding of the article as you review the translation.
- Compare your own summary of a paper with one written by AI. AI can be a useful tool for providing a summary and supporting your reading of academic papers. Comparing your own understanding of the paper with an AI output can be a useful approach to developing your critical reading skills – both by recognising things you may have missed, and by giving you an opportunity to critique the AI output.
- Critically review all AI outputs. Ensure you critically review all AI outputs for accuracy to support your learning, and verify the outputs against other sources, e.g. can you locate all references and are they accurate to the text, and are the definitions correct?
Suggestions of prompts to try:
- You are an expert in [subject] and an educator who is good at giving great explanations to beginners. Make a table of the key concepts needed to understand the content of this paper.
- Give me a list of 20 key terms in this paper and break it into five categories.
- Rephrase this definition as a list of bullet points to help me understand it step by step.
- Make a list of propositions in this text in the format “X is a type of Y”, “W is caused by X”, “A explains B”. Put it into a table with three columns.
- Rephrase this sentence in simpler language that a non-expert can understand. You can break it into multiple sentences, if needed.
Five ideas for academic writing and presentation skills
Generative AI tools can be useful in developing your academic writing skills and providing initial feedback on them, translating between different styles and critiquing writing. AI tools cannot replace the need for you to develop these skills through teaching and independent learning.
Here are five ways you can use AI to develop and get feedback on your academic writing, suggestions for how you might provide useful prompts, what some of the limitations are of the AI generated outputs, and how to engage critically with them to augment your learning.
- Examples of writing in different styles and genres. AI can provide examples of many different written outputs. You may find this useful in identifying different styles of writing, appropriate to different tasks.
- Feedback on your writing. AI can provide rapid feedback on your writing, and this can be helpful for improving it, e.g. in relation to grammar and structure. AI cannot provide feedback nuanced to the rigour and expectations of academic writing in your specific subject so it cannot replace tutors’ feedback grounded in the context of your discipline. Be sure that you proof-read your own work, as this is an essential skill in academic writing, and do not use generative AI tools to make material changes to work in draft.
- Help you get started in writing. AI can be useful in overcoming writer’s block by providing some inspiration or points to consider when you are about to start writing.
- Suggesting ideas for graphics, images and visuals. AI can be useful in suggesting how you might present information in graphics, images and visuals to move beyond text-based presentation of information.
- Critically review all AI outputs. Ensure you critically review all AI outputs for accuracy and verify the outputs against other sources, e.g. can you locate all references and are they accurate to the text, and are the definitions accurate?
Suggestions of prompts to try:
- Here is a paragraph I wrote. Rewrite it as a list of bullet points with the key arguments I made.
- I am a student of [subject]. I wrote this as an example of academic writing. Give me feedback on where I could improve. Focus on clarity, academic language and grammar.
- Suggest some practice exercises I can use to improve my weak points.
Five ideas for supporting your learning
Generative AI tools can be useful in supporting your academic studies.
Here are five ways you can use AI to support your studies, suggestions for how you might provide useful prompts and what some of the limitations are of the AI generated outputs. Be sure to verify any AI outputs against other established sources to ensure their accuracy, and cross-check to confirm your understanding.
- Preparing for lectures. You can ask for key concepts related to the topic of the lecture and use the lecture to compare this with your initial understanding.
- Engaging with new or complex topics. You can ask AI for explanations to help you develop a better understanding for yourself. For example, you can ask for alternative explanations of a topic or analogies from different perspectives.
- Organising your notes. Generative AI tools can be used to convert your notes into structures that are easier for you to review. You can ask for a table of key concepts, facts or figures organised by different categories. Timelines, workflow summaries, outlines you can convert into mind maps or mnemonic devices can be generated to help you remember facts. Remember, even if AI looks like it is copying things from your text, it is actually generating the new version from scratch using your text as context, so you must review all outputs for accuracy.
- Enhancing your language study. You can use AI to improve your language learning. You can ask to have a conversation at a certain level on a specific subject, receive feedback on your conversation, or generate sample texts for practice. Always ensure you cross-check the outputs against other sources as outputs in some languages may contain basic grammatical errors. If working with long texts, for languages other than English, the volume of text you can translate or work with will be much shorter. Some languages with non-Latin alphabet (like Arabic) may only be able to fit 1/5th or even less than a language like French. Many languages not well represented on the internet may only work very partially with generative AI.
- Developing your coding skills. You can use AI to develop your coding skills. Write code as you do normally – writing in (functional) chunks and testing the chunks for the right functionality yourself. You could then compare these chunks with a given AI tool. You may find it useful to ask your tutor which AI tools are more appropriate for coding. As with all AI outputs it is not perfect, and any code needs to be run to check that it functions as expected. Common errors are: the code may produce an error message, the description of the code or suggestions for improvement are inaccurate, the code is completely incorrect, or uses features not present in the language or the code generates references to outdated or non-existent libraries. It would not be appropriate to use AI to write a code, or to support the writing of a code, on which your coding skills will be assessed.
Suggestions of prompts to try:
- I’m going to a lecture on [subject]. Give me a list of 20 key concepts I need to be prepared for. Break them into five categories and explain how they relate to each other. You are an expert in [subject] and always explain things in a way that is easy to understand for a student at [level]. Make a table of these terms with four columns. 1. Term, 2. Definition, 3. Category, 4. Related terms. Make a series of statements using these terms that will reveal the relationships between them. For example, working memory is a type of ….
- As an expert in [subject]. Here’s an explanation I was given of [concept]. Can you suggest some alternative explanations of [concept] to help me learn. I am a [level] student at university and have already taken [classes in…].
- You are an expert in [subject]. Here is a definition I was given of [term]. Can you give me five concrete examples that illustrate the definition. I want to be able to apply them in practice. For each example, specify exactly how it illustrates the definition.
- Here are my notes from class. Make me flashcards to help me study. Each flashcard should have the term on one side, and a brief explanation on the other. Use my explanations where available. If you have suggestions for alternatives, put them in italics so that I know what is in my words.
- You are a shop assistant in a bakery in Germany. I am a beginner student and I’ve come in to buy some bread. Let’s have a conversation in German. You start by greeting me and asking me what I want. You will pause and give me a chance to answer. Use simple language and always respond even if I say something using the wrong term. Do not use any English until I tell you to. Then make a table of all the things I said with three columns: 1. What I said, 2. What I should have said in correct German. 3. What error I made and how to correct it.
- I want to write code to do [describe task]. As an experienced programmer and a coach of new coders, suggest the best way to help me get started. Suggest the best language, coding environment, and dependencies. My level is [describe level].
Five tips for selecting the right generative AI tools for the task
There is a huge range of generative AI tools available, e.g. Microsoft Bing Chat, ChatGPT, Google Bard, Claude. Some of the tools are freely available and some with costs that bring extra features such as being able to work with large volumes of text.
- You will find it useful to try a few different AI tools and be aware that different tools will give different outputs using the same prompts and a tool will also generate different outputs to the same prompt.
- You will get different responses using the same prompts from the same AI tool. AI outputs are not repeatable, and all tools can generate outputs that can contain inaccuracies and fabrications.
- You could spend a lot of your study time trying out different AI tools. Be careful to manage your exploration of tools alongside managing your time.
- AI tools may draw on data that can be months or years out of date and whilst outputs seem plausible, they may contain errors and/or reflect biases from the original data, e.g. Western perspectives are overly represented.
- AI tools will not replace the need for you to develop your own knowledge and skills as an independent learner.