Technology enhanced by artificial intelligence (AI) has been present and embedded in our education practices for quite some time now. From spell checkers, text prediction in Microsoft Word, Grammarly, to Turnitin, educators and students in higher education have successfully engaged with and adapted to previous AI technologies. The appearance and growing popularity of Chat GPT in 2022 prompted wide discussions on the use of Language Learning Models (such as Chat GPT) and the multiplicity of research apps and platforms which incorporate AI for use in education.
As a result of the queries we have received on this topic, we have decided to compile this LibGuide to provide general guidance on how AI might be ethically incorporated into the research process. In the sections below, we will draw attention to some points to consider when using AI-enhanced technology, highlight the University's position on the use of AI and academic integrity, and introduce some tools that can allow you to incorporate this new technology ethically.
Please note: This LibGuide is a work in progress, and we will update it as new developments occur. If you have any suggestions or would like to draw our attention to something we may have missed, please email the Academic Skills Librarian.
Would you like to take part in a focus group exploring the use of generative AI in academic studies? If so, please complete this short form to share your thoughts and register interest in participating in a focus group: Do you use generative AI for your academic work? Participation is open to all students at the University of Cambridge.
Over the years, a variety of definitions have surfaced for AI. Below are some definitions that can help give a sense of what is meant by the term artificial intelligence:
"At its simplest form, artificial intelligence is a field, which combines computer science and robust datasets, to enable problem-solving. It also encompasses sub-fields of machine learning and deep learning, which are frequently mentioned in conjunction with artificial intelligence" (www.ibm.com/topics/artificial-intelligence).
"Artificial Intelligence (AI) is a rapidly evolving field of developing machine intelligence to replicate, and in some cases, exceed human cognitive capacities. The most prominent techniques in AI that have risen to wider attention are machine learning, large language models, and natural language processing (generative AI) which, when prompted by user input, can produce seemingly intelligent responses replicating that of human interaction" (https://blendedlearning.cam.ac.uk/guidance-support/ai-and-education).
From these definitions it becomes clear that when we are referring to Artificial Intelligence, we are referring to a field of study that is concerned with using computer science and various forms of machine learning to problem solve and provide responses to problems in a similar manner as humans. This is the type of technology which is already present in spell checkers, predictive text in Microsoft Word, and proofreading and grammar checkers like Grammarly.
Image created by BING Copilot in response to the prompt: Create an image of AI as an embodied thing.
Generative Artificial Intelligence is a type of machine learning that has prompted the recent discussion on the use of artificial intelligence in education. It is a type of deep-learning model that can take existing raw data, such as the entirety of Wikipedia, and learn to create new, unseen data, such as essays or images when prompted. Generative AI learns through solving problems and receiving feedback, and it is for this reason that you may often hear about the importance of creating clear, precise prompts when engaging with Chat GPT or similar applications.
Image created by BING Copilot in response to the prompt: Create an image of generative AI as an embodied thing.
Machine Learning - Is broadly defined as a "subcategory of AI [that] uses algorithms to automatically learn insights and recognize patterns from data, applying that learning to make increasingly better decisions." (https://ai.engineering.columbia.edu/ai-vs-machine-learning/). Machine learning can be broken down into further categories: supervised; unsupervised and reinforcement.
Natural Language Processing is an "interdisciplinary field that lies at the intersection of linguistics, computer science, and ML [Machine Learning]" (Surdeanu & Valenzuela-Escárcega, 2024). It can be found in a wide array tools, programs and software we use in our daily lives, such as the use of search engines on the internet, social media, and organising our emails. The goal of natural language processing "is to get computers to perform useful tasks involving human language, tasks like enabling human-machine communication, improving human-human communication, or simply doing useful processing of text or speech" (Jurafsky & Martin, 2014).
While we may already by comfortable with many forms of AI in our day to day life, below are some points to consider as Generative AI continues to create new possibilities to influence how we engage with learning and our own research.
Be wary of false citations, even if the sources sound legitimate or you are provided with quotes from the article. Large Language Models (LLMs) like Chat GPT can often "hallucinate" or invent sources that sound convincingly real. It can make up articles from established publications and even court cases with quotes from imaginary opinions.
Reading summaries is not a replacement for critically engaging with the text. Think of AI software that provides summaries of text as your first scan of a reading; it can help you get a general sense of the reading, but you may miss out on a lot of the nuance and subtle features of the text that can be essential for using it later in your own writing.
Be aware of the possibility of bias in the information you are presented. Since generative AI learns from existing raw data, it may, at times, provide responses that take data out of context or reinforce harmful stereotypes. Though the responses provided by generative AI may give the appearance of a complex, reflective, and deeply reasoned process, remember, current platforms are heavily susceptible to the quality of prompts that are entered and lack of moral understanding of the information they provide. Having a critical approach to the information you are presented is just as necessary as if you were engaging with other resources for your research.
Don't let generative AI do your work for you. Relying on generative AI to substitute the work needed to develop a skill or practice is not only a detriment to your own learning experience, but it offloads, as Marc Watkins notes, "the entire moral, ethical, and responsible thinking we expect from a human being."
Last but not least, while generative AI can help you work through various stages of the research process, remember you are ultimately responsible for knowing what is needed at each stage of your research and for the information you submit for evaluation.
As of February 2024, the University of Cambridge has issued the following regarding the use of AI:
"Students are permitted to make appropriate use of artificial intelligence tools to support their personal study, research and formative work. Where doing so, it is recommended that you discuss this with your supervisor or lecturer to understand how best to engage with these tools whilst still benefiting from the educational experience as intended.
A student using any unacknowledged content generated by artificial intelligence within a summative assessment as though it is their own work constitutes academic misconduct, unless explicitly stated otherwise in the assessment brief.
We encourage staff to clearly communicate their expectations to students and encourage use of available guidance where relevant and useful. If you have any concerns regarding the potential use of artificial intelligence, please discuss this with your supervisor or lecturer to ensure you have the most relevant and up-to-date information."
For further information, please visit the University's Artificial Intelligence section on the Academic Misconduct and Plagiarism page.
Most generative AI tools will, by default, collect the data you enter in your prompts to help train and develop the AI tool further. In many cases you will have the ability to option to opt out of allowing the company to use your data this way, but you may need to actively select this option.
If you are entering research that you are not ready to share publicly, please be aware that unless you are able to opt out of allowing your information to be used for training, it may reappear in the responses the AI tool provides to other similar prompts.
If you are using a generative AI tool to help transcribe interviews or analyse participant data, you will need to be aware of how it stores and manages your data. Improper use of generative AI for these types of tasks may constitute a violation of ethics for your research project.
Copyright gives authors a legal right to their own intellectual property (IP), so that they own it as they would a piece of physical property. Other people are not allowed to use it for a specific period, unless they are licensed to do so by the author.
Generative AI models like LLMs need to be trained on large datasets. These are usually culled from a variety of sources, many of which may not be owned by the company producing the model. Data entered into the model as prompts and the content produced in response to those prompts is also fed back into the system. This raises several questions, most importantly:
There are now ongoing lawsuits from companies seeking compensation for the use of their material to train AI models, including The New York Times vs OpenAI and Getty Images vs Stability AI. The Authors Guild has also filed a class action lawsuit against OpenAI and Microsoft, following an open letter to the leaders of AI companies in which they demanded that these companies obtain permission for use of copyrighted material in generative AI models and compensate writers for use of their work in this way, regardless of current legislation.
The outcomes of these cases are likely to provide some clarity. Governments are also looking to produce guidelines and legislation in this area. For now, the important thing for researchers is to acknowledge their use of AI tools as they would other sources, to avoid plagiarism. You should also look to keep up to date with future developments.
As the capabilities of generative AI continue to evolve, it will be important to have a sense of each tools limitations and how you can use it without breaching the guidelines of academic integrity. Below are some ways you can use various platforms to assist you in your study without participating in academic misconduct:
If you use generative AI to assist your study, remember that it should not be the only method used to study. You will have greater success if you incorporate its usage alongside other study techniques where you can develop your subject knowledge and critical thinking skills.
Image created using DreamStudio in response to the prompt: Create an image of AI in the style of Rodin's "The Thinker".
Generative AI offers students powerful tools to brainstorm, organize, and even generate text. However, it's crucial to remember that AI is a tool, not a substitute for critical thinking and academic growth. Developing strong academic skills, such as defining your information need, critical reading, note making, and academic writing, is essential not only for building a strong academic skills foundation but for developing a critical mindset to effectively utilise generative AI.
Below we have put together a sample checklist you should consider if you are allowed to use AI in your coursework. The checklist is based on Elon University's The essential AI "how to" Manual. You can also download a PDF version of this checklist with further suggestions of how to integrate generative AI into your research process in a more meaningful way at the bottom of this section.
Remember, AI is a tool to enhance your research, not replace it. If you need support in developing the skills necessary to effectively engage with generative AI, explore the resources on our Wolfson College LibGuide, including videos of our workshops, materials, and activities to help you build a strong academic skills foundation.
The first step to using generative AI platforms effectively is to create clear, precise prompts that will allow the platform enough information to produce the best answer for your query. Below are some top tips from Monash University on creating better prompts:
Sometimes it can be helpful to have a framework to help guide you as you create prompts. Leo S. Lo provides 5 principles to follow in the CLEAR framework for prompt engineering:
Below, you will find a list of tools incorporating AI for academic purposes. We have divided them based on each tool's emphasis and placed them in the relevant stage of the research process. Where possible, we have also linked to videos introducing the tool and providing guidance on using it. This is by no means an exhaustive list of what is available, so if we have missed an important tool, or you would like to let us know of one that you use and you have found helpful, please feel free to let us know using the email link at the top of the page.
is an academic search engine that uses AI algorithms to search a document and recommend relevant research papers. You can upload papers through its website or you can add it as a plug-in for Microsoft Word or Google Docs.
is an academic search engine that can help you build a network of papers from citations using its algorithms to find similar papers and it offers a Literature Connector that is meant to help you bridge
is an AI research assistant that can find specific research articles that cover a general topic or answer a specific question. You can also read summaries of the abstracts your search brings up and search forward and backwards using the citation graph.
Create an interactive literature map using collections of articles you input and articles it discovers by searching academic databases. You can use Litmaps by creating a free account, and you have the option to upgrade to a paid version..
Creates citation maps and networks of recommended articles based on articles you have provided. You can set up alerts for new literature related to your research, integrate your collections with Zotero, and collaborate with others to curate your collection.
promises to help you absorb and retain course readings and brings more structure to your study. You can create summary cards to help you engage with the reading, begin your pre-writing or even begin structuring your revision notes. Scholarcy offer a free and paid version of their service.
allows you to upload papers and ask questions to its AI Copilot to explain difficult passages. SciSpace also has a Google Chrome extension so you can use its Copilot on webpages and articles you have access to through your institution.
allows you to upload PDFs where you can read, organise, find and share your research. It allows you to create "Concepts" in your workspace similar to a synthesis matrix and suggests future papers based on your uploads.
is note making and productivity software that offers a variety of organizational tools to assist in time management, organizing your notes, and project management. It provides free templates and allows you collaborate with others.
is free software that links your ideas and makes visualisations to help you naviagate between them. It is also extensible so that you can tailor it to fit your purposes. It lives in a local folder and not the cloud, so you can keep control of your notes. It will take a bit of getting used to as it uses Markdown format of plain text files but this is to make your notes futureproof so you can move your notes to another editor.
is an experimental software by Google labs which incorporates Google's Gemini 1.5 LLM to help with your note making. Key features of the software is its ability to provide summaries to texts you have entered into your online notebook, creates flashcards, and even create a podcast. "A key difference between NotebookLM and traditional AI chatbots is that NotebookLM lets you “ground” the language model in your notes and sources"(https://blog.google/technology/ai/notebooklm-google-ai/).
Get language feedback, explore the writing patterns of journal articles, browse academic phrases to use, automatically paraphrase text, and auto-generate your title.
Offers support in checking grammar and writing style. In addition, Grmmarly has also started to use generative AI to help the user write, rewrite, ideate, and reply with simple prompts. You will need to subscribe to a plan with Grammarly to get a monthly allowance of prompts to begin writing with generative AI assistance.
Think Machine is a mind mapping programme with built in AI to help you visualize your mind maps in 2D and 3D. You can also use this programme to help you brainstorm ideas and see connections between your ideas.
Inspired by Ernest Hemingway's style of prose, this editor also you to copy and paste your text into its editor software and receive feedback on sentence construction, overuse of weak adverbs, excessive passive voice and word choice. Each category is colour coded and you can see the reading difficulty of your text. The video below is for the paid version Hemingway Editor Plus, but if you would like to see how the free version works please see: How to use Hemingway Editor 2024.
uses AI to generate working presentations, documents, or webpages that users can customize and refine. Gamma also has a library of templates to help kickstart your creativity.
Powered by Stable Diffusion, a generative AI model that can create realistic images, art and animation from text and image prompts.
The latest version of Open AI's DALL-E text-to-image generation model, this version is only available with a subscription to Chat GPT Plus. In the video tutorial below you can learn how you can access DALL-E 3's features on Microsoft Bing's Copilot and some of the difference between DALL-E 3 in Chat GPT Plus and the version available in Copilot.
Watkins, Marc. 2023. “Teaching Is Not a Problem for AI to Solve.” Substack newsletter. Rhetorica (blog). November 30, 2023. https://marcwatkins.substack.com/p/teaching-is-not-a-problem-for-ai.
Unless otherwise stated, this work is licenced under a CC-BY-NC-SA 4.0 licence by Wolfson College Cambridge.
