Describe and Document AI Use

Reading Time: 9 minutes

Disclosure of AI use involves sharing what tools were used, how they were used, and why they were used. They may also include human effort or a description of how or why they were not used. AI disclosure in scholarly activity promotes transparency and the thoughtful, responsible use of AI.

Conestoga Guidelines for Student AI Use

You may wish to provide students with the optional use of generative AI when completing learning tasks and/or assignments. For assignments, Conestoga’s Evolving AI Guidelines provide the following recommendations to faculty:

“Written instructions must clearly state whether use of AI is permitted [for assignment]. When AI is permitted, students should also cite and reference its use appropriately. Best practices indicate they should also include an acknowledgement. For example, an assignment completed in collaboration with Copilot should include a statement like “This assignment was completed with the assistance of MS Copilot (February 20, 2024 version).”

Create and share a Statement of AI Use that explains your expectations to students up front. Note that how and to what extent you ask students to use and disclose their AI use will depend on many factors.

Types of Disclosure for Different Purposes

The type of AI disclosure will depend on the goal and the learning task’s process. AI use should never be used to replace or shortcut the demonstration of knowledge and/or skills.

Below are 5 ways to disclose AI use.

1. APA-style citations and references

AI can be an information or research resource (Weaver, 2024). Citing an AI chatbot involves acknowledging the source of information obtained from the chatbot in your work. It gives credit to the creators of the AI and for allowing others to trace the origin of the information.

When to use it: When quoting, paraphrasing, or summarizing AI as an information source. In essence, the creator of the AI is the “author” of the source from which quotes, paraphrases, or summaries are used.

Share your expectations: Explain that AI chatbots create “new” outputs every time, derived through a predictive algorithm that draws from the LLMs training data. If you use the words or ideas from the AI output, you must cite and reference it for academic integrity. See the Hub post Citing Generative AI for more information. Help students learn about AI citations by showing them the Generative AI Toolkit for Students – How to Cite AI.

2. Describe AI collaboration

AI can be an educational resource for conceptualization, information collection, visualization, and writing review and editing (Weaver, 2024). Describing AI collaboration involves summarizing the process by which AI was used to design, develop, or refine the learning task output. This describes how, how much, and when AI was used.

When to expect it: When students use AI for brainstorming, researching, outline, categorizing, analyzing, evaluating or creating a product. In essence, AI is an aid or collaborator in the work.

Share your expectations: Explain the permitted ways that AI may be used, then give specific examples of appropriate summaries that give sufficient levels of detail. Here are some examples.

1. Initial draft generation: “I used generative AI to generate the initial text draft. I then reviewed, edited, and refined this draft to produce the final output.”
2. Idea generation: “I used generative AI to generate various ideas and concepts. I selected the most promising ones and developed them further on my own to create the final text.”
3. Content expansion: “I provided the AI with a basic outline or a set of key points. The AI then expanded on these points to create a more detailed and comprehensive text, which I revised and finalized.”
4. Iterative refinement: “I used the AI in an iterative process. I would input a draft, the AI would revise and enhance it, and then I would review and adjust the AI’s output. This cycle was repeated until the final text was produced.”
5. Language enhancement: “I wrote the initial text, and then used the AI to enhance the language, improve the style, and correct grammatical errors. The final output was a polished and refined version of my original text.”

You might create a list of appropriate options for students to select and adapt. You might also wish to to ask students to include this description at the beginning or end of their submission, or in a message along with their submission in eConestoga.

3. Provide AI Use Documentation

Providing AI documentation involves including artifacts of the outputs or conversations that AI provided for students.

Expect it: Students who use AI to create specific outputs are asked to “do more” with it (critique, evaluate, revise, verify, etc.). Or, students are asked to use AI as a conversation or critical thinking partner. In essence, there is a need to verify that the learning task output differs from what AI-generated or that a “conversation” was had between AI and the student.

Share your expectations: Students can document their use of generative AI in several ways to demonstrate their active role in the development process. Here are some examples.

1. Screenshots: Capture images of the genAI interface during use, showing the input given and the AI’s output.
2. Prompt and Output Logs: Record and share the interactions with the genAI, including the prompts or questions, the AI’s responses, and adjustments made based on those responses.
3. Annotations: Students may be asked to highlight or annotate their final submission, indicating which parts were generated or influenced by the AI and providing a brief explanation.
4. Process documentation: Create a step-by-step account of the assignment process, detailing how the AI was used at each stage. See below for an example.

Note that MS Copilot only allows individual prompts and responses to be copied. You may need to give instructions and/or demonstrate how students should keep track of their AI use. For instance, students may not know how to take screenshots.

4. Complete an AI process form or reflection

Completing an AI disclosure or process form helps students explain how much AI helped their work. It covers idea creation, research, planning, writing, and editing. It clearly distinguishes how AI may be used (or not if students opt out of AI use). This practice makes the role of AI transparent, helps students understand their learning, and makes them think about the correct use of AI for the task(s). It may be a step that is added to requiring students to cite and reference AI outputs.

When to use it: Use this strategy when AI is being integrated in multiple ways throughout an assignment or to distinguish levels of permitted and prohibited uses of AI.

Share your expectations: If you choose to use a process form along with an assignment, you may wish to review it with students well before the due date. Also, ensure that students are clear about your expectations for the permitted (and prohibited) use of AI and how they should keep track of their AI use. See The Optional Use of Generative Artificial Intelligence (GenAI) in Assessments for details.

5. AI Acknowledgement Statement

An AI Acknowledgement statement precedes a learning task output or assignment, and it identifies the strengths, limitations, and risks of AI in general and how AI was used specifically.

When to use it: Before a large project, presentation, or work that requires acknowledging important information for an audience. Or to anticipate questions or concerns the audience may have about general or specific potential harms of AI.

Provide a model or an example for students to follow. Talk about what it discloses generally (harms of AI) and specifically (how AI was used, especially to minimize harms).

How might AI disclosure strategies be encouraged?

Since AI disclosure is a new and additional task for students, they will likely need support and encouragement to be successful with AI disclosure activities. See the accordion below for ideas supporting AI disclosure in learning tasks and assignments.

When might AI documentation not be necessary or desirable?

It depends. Disclosure may not be necessary for tasks that students already know how to do and/or are automated to save time. Or, you may only be interested in the “product” of their learning, not the “process.” For example, you may decide that students who use AI to brainstorm or acquire resources for a class activity or assessment may not need a description/documentation of their use of AI.

The norms of the academic discipline of your professional area may also guide your decisions. When in doubt, refer to Conestoga’s Evolving AI Guidelines and contact your supervisor.

Final thoughts

If you wish to provide students with optional use of generative AI for activities and assignments, discussing the risks of using 3rd party genAI tools in class is strongly recommended. Also, you can help students understand your expectations by demonstrating what you ask of students. Finally, please note that students cannot be required to use a specific AI tool for class activities and assignments. Alternative opportunities with equitable support must be provided for any assigned tasks.

References and AI Disclosure

Queensland U (2024). Generative AI tools for assignments. Library Guides – Referencing.

Weaver, D. (2024). The artificial intelligence disclosure (AID) framework: An introduction. University of Waterloo.

Note: In this post, examples were developed in collaboration with AI: detailed prompts were provided, and outputs were human-revised and finalized.