Citing and Acknowledging AI Tools in Research

Artificial intelligence (or AI) tools are impacting every aspect of college life, and CMU Libraries are no exception. Many library resources, including the catalog (Smart Search), are increasingly incorporating AI either by default or as opt-in research assistants. There are also many free tools out there (such as Semantic Scholar, Scite, and Consensus) that are built to search literature and summarize results. Frequently used generative AI tools (such as ChatGPT, Claude, Google Gemini, and Microsoft Copilot) can also help discover scholarly literature, if prompted to do so. 

The use of AI in academic research and writing is a complex topic with no simple answers. However, librarians are leaning into their expertise on the ethical use of information and providing some guidance for those using these new tools to help in their academic work. 

Whether folks choose to use AI tools or not, two key points of academic honesty are avoiding plagiarism and being transparent about the research process. Both of these can be accomplished when using AI for sourcing literature. 

According to guidance from major citation styles, any text in your work that was generated with AI should be cited both in-text and on the references list. It's generally recommended to include the prompt you used in the text or in the in-text citation. For example:  

When prompted to "please explain how regret features in Robert Frost's The Road Not Taken", the text generated by Copilot suggested that regret wasn't caused so much by choosing the wrong path, but rather never knowing what the other might have brought (Microsoft, 2026).  

The reference list for the paper would then include: 

Microsoft. (2026, January 9). Regret in Frost's "The Road Not Taken" [Generative AI chat]. Copilot. https://copilot.microsoft.com/chats/HUYedj5AQrT6uBBEmNTyJ  

This example models the style offered by APA, but similar examples are given by MLA, Chicago, and others. 

In addition to citing any content generated by AI, it is ethical practice to acknowledge and disclose all uses of AI in the research and writing process. This is already the norm of large academic publishers, who provide specific guidance to authors on what to include in such an acknowledgement. These may be the name, publisher, version, and link to the AI tools used; the dates the tools were used; a description of how the tools were used and why; and what the tools were NOT used to do. A further option for an AI acknowledgement, as noted by Weaver in her 2024 Artificial Intelligence Disclosure (AID) Framework, is to list out every stage of the research and writing process and describe how AI tools were used in each. Another practice in use for academic work is to require students to include both the full text of the prompts used and original outputs generated with those prompts. 

For example, an acknowledgment of AI use in a class paper might look like this: 

Between January 4th and 9th, 2026, I used the CMU institutional version of Microsoft Copilot in my research and writing process as follows. 

Conceptualization: I used Copilot to help me refine and better state my research question. 

Literature Review: I asked Copilot to search for relevant literature on my topic. I independently checked the sources in Google Scholar and the library catalog to be sure they were real, and that the full text contained the information Copilot suggested they did. I also used Copilot to help me change my list of sources into an APA-style bibliography. 

Writing: After asking Copilot to create an outline, I wrote the original draft of my paper myself. I then used Copilot to check my grammar and to make my phrasing clearer and more focused. 

All my original prompts and outputs are attached to this assignment. 

Individual instructors may have different requirements for how they want their students to use, or not use, AI tools, and how to acknowledge those uses. However, when conducting a research project, tracking all AI use with dates, prompts, and reasons is a great practice to maintain, ensuring all your bases are covered. 

Though generative AI and AI research tools are new and constantly evolving, being clear and honest about citing work not your own is a long-standing practice that has not changed. As CMU writers adapt to these new tools, the Libraries are here to help them use AI effectively and ethically.