Generative AI: Citations and Disclosure Statements

In general, you are all aware that you cite sources to avoid plagiarism. Because AI generated content is not coming from any single source, and because the training data cannot be attributed to any specific author, you cannot cite AI in the same way you might cite a scholarly article or website. This is why many academics and researchers are encouraging the practice of disclosure. Kari D. Weaver provides excellent rationale on the way in which generative-AI tools are affecting the traditional citation practices of higher education in her article "The Artificial Intelligence Disclosure (AID) Framework: An Introduction".

An excerpt...

"Within the contexts of education and research, and particularly within higher education, the citation has long been the standard tool for providing transparency and connection in the transfer of ideas across scholars, framing of arguments, and design of methodologies.

Unfortunately, citations do not fully meet the needs of today’s AI-enabled world. Citations emphasize the fixed form of a tangible output. This is incongruent with today’s generative AI systems, where the specific interplay of prompt, model, and model parameters creates a unique output that is  not always repeatable, reproducible, or recallable, depending on the technology. Citations also focus on the ideas posed by an author, whereas generative AI can serve a variety of meaningful functions in the writing process, including researcher, editor, critic, collaborator, and more. While today’s citation practices do help provide some transparency, they are not sufficient, in and of themselves, to capture the varied ways AI tools function or are being used across contexts (Weaver 2024)."

Please note this example is a very detailed AI disclosure statement. Professors may require less or more specific notes about your generative-AI use for course work. Best practice is to always check with your professor before using any generative-AI tools.

Scenario

Fatima’s professor allows the use of AI tools but requires students to cite their usage and be transparent about how they employed AI in their research process. Here’s how she incorporates AI.

AI Usage

1. Literature Review: Fatima uses an AI research assistant to help find relevant peer-reviewed articles. She documents the AI tool used and the search parameters she set.
2. Summarization: Fatima employs an AI tool to generate concise summaries of key points from each article. She reviews these summaries critically and uses them as a starting point for deeper reading.
3. Outline Generation: Using an AI writing assistant, she creates an initial outline for her paper based on the main themes she's identified. She refines this outline manually, adding her own insights and restructuring where necessary.
4. Draft Writing: Fatima uses AI to generate rough drafts of certain sections, which she then heavily edits and expands upon with her own analysis. She keeps track of which sections were initially AI-generated and how much she modified them.
5. Citation: She utilizes an AI-powered citation tool to ensure proper formatting of her references.

Citation and Transparency

Fatima includes an appendix in her paper titled "AI Tools Used in Research Process." Here, she details:

1. The specific AI tools used (e.g., ChatGPT, Grammarly, Google Scholar AI)
2. How each tool was employed in her research and writing process
3. The extent of AI involvement in each section of the paper
4. Her process for verifying and expanding upon AI-generated content

Fatima also includes citations for the AI tools themselves, treating them as sources that contributed to her research process.

This example was originally developed by OSU librarian, Chris Rosser using Claude AI.