How to Cite or Acknowledge AI Tools as a Source or Author in Academic Writing: ChatGPT and Other Tools

NOTE: There are two more recent and detailed posts about this, guest-written by Dr. Kristin Terrill of Iowa State University. The first is “Should I Cite AI Tools,” and the second is “How to Cite AI Tools.”


Have you been using ChatGPT or other AI-based tools to help you do academic research and writing? If so, are you wondering about how it should be cited? Well, you’re not alone. This blog post helps you navigate this situation with integrity by explaining how to approach citing and/or acknowledging all kinds of AI-based tools.

Academics are increasingly integrating various forms of generative artificial intelligence (AI) tools into their research and writing workflows, which creates both stimulating opportunities and challenges. At Moxie, we are often approached with a critical question: How do we attribute AI tools (including Moxie tools) in academic writing, especially following style guides such as APA and MLA?

What?

First, let’s explore what these tools are and how they function. Most AI tools, including Moxie’s tools, are layered on top of a large language model (LLM). For example, many of our tools are powered by GPT-4’s capabilities with an additional application layer. Tools like Perplexity.ai are also built on top of an LLM. If you try to find out how to cite these tools, you might come up short because while the APA and MLA blogs provide information on citing directly from an LLM (e.g., ChatGPT, Claude, Bard, etc.), they do not explain how to cite tools built with application layers. So, what you need is a more nuanced approach to citation, which adheres to the principles of transparency and academic integrity.

When/Where?

Since we primarily use APA 7, we’ll start there, but you can apply these suggestions, to whatever style guide you use. Drawing parallels with APA’s recommendations for citing large language models like ChatGPT, this blog proposes a citation format for both Moxie tools and similar application-layer programs to be cited in a manner that acknowledges both the underlying AI technology and the specific application layer they provide. This can be extended to other tools developed by companies who, like us, are layering their products on top of an LLM. The caveat, however, is that this is just a proposal.

At the risk of being repetitive, it’s important to emphasize that there is already a guideline for citing an LLM directly. The suggestion below is for AI-based tools with an additional application layer that provides a unique functionality (e.g., doing some of the prompting for the user). Therefore, the citation should not only acknowledge the underlying language model but also the specific tool and its unique features.

  1. Identify the Tool and Version: Clearly specify the name of the tool and the version used, if applicable. This is like citing a specific version of software.

  2. Describe the Use in your Paper: Decide on acknowledgment vs. citation —

  • If you used the tool like a tutor, editor, or coach, then you would acknowledge it (rather than cite it) in much the same way as if you had worked with a human editor, coach, or tutor. (Or maybe not? Do you acknowledge your use of MS Word’s spelling and grammar checker? Grammarly? — these are nascent issues that don’t have solid answers to be found). When writers acknowledge the support of a mentor of any kind, they usually do it at the beginning or end of a manuscript according to established practices by the journal (in the case of an article) or the university or professor (in the case of dissertations, theses, coursework). Another appropriate location for this is in the cover letter that accompanies the manuscript. A recent journal article in Frontiers in Artificial Intelligence provides specific language you can use as a template.

  • For another take on this topic, please see chapter 13 (Acknowledging and Citing Generative AI in Academic Work) in the e-book, Write What Matters.

An article that provides suggested wording for acknowledging AI use in academic manuscripts.

  • If, on the other hand, the tool was used for generating text (e.g., summarizing, paraphrasing, examining themes, eliminating repetitions/duplications, etc.), then it should be cited (as opposed to acknowledged). This might be placed in the Methods section, or, depending on the exact use/context, you might put it elsewhere. Please note: It’s very context-specific, and there’s no one-size-fits-all solution. Wherever you decide to put it, we suggest you describe how the tool was used in the research process and then cite the tool as software. Since AI-based tools use an application layer on top of a language model, they should cited similarly, with the developer or organization responsible for the tool as the author, as shown below. The rationale for including the application layer in the citation is that it differentiates the specific functionalities of the tool from the underlying language model.

  • Provide Access Information: If possible, include a URL or another form of access information where the tool can be found or used. For Academic Insight Lab and other paid programs, you will need to use the URL of the company (e.g., www.moxielearn.ai) and not the exact URL of the tool, as it is likely behind a paywall that you access via a username and password.

How?

Here is a general template for citing a tool that uses an application layer (e.g., Moxie’s tools, or programs such as Perplexity, Jenni, ChatPDF, etc. in APA 7 style:

Reference List Entry:

Developer or Organization Name. (Year). Tool Name (Version) [Application layer over GPT-4]. URL

In-Text Citation:

  • (Parenthetical citation: Developer or Organization Name, Year)

  • Narrative citation: Developer or Organization Name (Year)

Example of Citing a Moxie Tool

Let’s say the Moxie tool is called “InsightGenerator,” and it’s built on top of GPT-4. Here’s how you might cite it:

Reference List Entry:

Moxie (2023). InsightGenerator (Version 1.0) [Application layer over GPT-4]. https://moxielearn.ai

In-Text Citation:

  • (Moxie, 2023) or

  • Moxie (2023)

Again, these are our suggestions only, and as there is not yet advice provided from prescriptive agencies such as APA or MLA, you might also consider confirming this is acceptable to your colleagues, collaborators, or supervisors. Regardless, by following these guidelines (or simply asking the question), you’ll help ensure that you are properly acknowledging the tools you use in your research, maintaining the integrity of your work, and allowing others to replicate your methods if desired.

A visual representation of citing AI tools in academic writing; how to cite chatgpt; how to use chatgpt for academic purposes

 

Why?

 Transparency in the use of AI tools is particularly crucial in the academic context for several reasons:

  1. Enhancing Research Quality: Transparency is vital for scholarship to uphold its core principles of ethical inquiry, rigorous analysis, and cooperative progress.

  2. Upholding Academic Standards: Legal and ethical standards are not just regulatory requirements but also the backbone of scholarly integrity. Transparency ensures that research complies with these standards.

  3. Fostering Academic Trust: Clear and obviated use of AI in research garners trust among peers, review boards, and the wider academic community. When the logic and process behind AI-generated data and analysis are open, it bolsters the credibility of the research.

  4. Promoting Collaborative Inquiry: Being candid about the AI processes you adopt facilitates shared understanding and cooperative effort, which is essential for interdisciplinary research and peer review. It can help you check your motivations for its use: Are you augmenting and supporting your research and writing workflows or just automating and taking shortcuts?

  5. Supporting Scholarly Development: Where learning and teaching are key goals, straightforward plain-speak about AI decision-making processes opens educational opportunities. It enables students and researchers to learn from AI methodologies and incorporate AI insights into their skill sets.

Takeaways

As academics, it’s essential to remember that our primary goal is to enrich academic discourse, not to replace the human element in our research. By citing AI tools appropriately, we uphold the values of academic rigor and transparency, ensuring that AI serves as a valuable ally for us. If you’d like to learn more, check out our YouTube channel for these tutorials:

Elevate Your Research with AI

Dive into a curated collection of AI tools specifically designed to enhance and streamline your academic journey.