»Æ¹Ï¶ÌÊÓƵ

Approaches to GenAI

Please note that the recommendations on this webpage are not meant to serve as policy and do not reflect the views of the university, individual departments, or faculty. Our recommendations are suggestions based on scholarship in Rhetoric and Composition and writing center pedagogy. Please consult university policy when handling any case of academic dishonesty.

Last updated August 2024

The CWE aims to be a site of collaborative learning that fosters a culture of excellence in writing, communication, and teaching. The recent interest in generative artificial intelligence (GenAI) technologies does not shift our primary goal, which is to promote learning and skill-building. Since OpenAI introduced ChatGPT in 2022, the conversation has focused on its potential and drawbacks. Many have noted how GenAI might enhance efficiency, productivity, and accessibility, while others have noted concerns about GenAI’s effects on , , the , and more. In the CWE, we aim to balance the potential of new technologies such as GenAI with awareness of their limitations, including how over-reliance on GenAI might have a negative impact on learning as well as problems with GenAI applications themselves, such as and .

Even though the conversation about GenAI shows no signs of slowing down, the CWE remains committed to its mission: to empower writers every step of the way. We do this by emphasizing the four goals we have for writing consultations: 1) making writers aware of writing as an iterative process; 2) helping writers understand writing and communication as rhetorical, as writers engage specific audiences to achieve particular purposes; 3) teaching writers to transfer their writing and communication skills to other situations, disciplines, and genres; and 4) assisting writers with locating and utilizing resources to support their work. 

In keeping with these goals, the CWE views the use of GenAI technologies as we do other software such as Grammarly or documentation software like EndNote: tools that can facilitate learning and skill development but should never replace them. For this reason, our approach to GenAI always begins by ensuring that writers first and foremost understand the task they are trying to complete. Whether they are learning to cite sources, craft thesis statements, or transfer writing skills between disciplines, our focus is always on supporting writers’ learning. Then, we can help students understand both the benefits and limitations to using GenAI to meet their goals, how their use of GenAI aligns with the university's academic honesty policy, and whether this use complies with policies regarding GenAI in the venue where they are submitting (for instance, course policies or journal requirements). For us, learning is at its best when approached as a social activity—when people come together to collaborate, share ideas, and learn from one another. Writing is also a social activity: whether readers are professors, journal editors, scientists, policy-makers, etc., they are human, and GenAI cannot duplicate the perspective of a human reader that the writing center's experts can provide.

For writers

• Should I tell the CWE if I used GenAI in my work?
  • Yes, it is helpful for us to know up front whether and the extent to which you used GenAI to compose your draft. This information will help us understand more about your goals as a writer and your writing process. We can also work with you to make sure the ways you’re using GenAI are allowed by the university, your professor, and/or external audiences.
• What will the CWE do if I bring work that has incorporated generative AI?
  • We will begin by having a conversation about your professor's policies (or, for other kinds of writing, such as peer-reviewed articles, the policies of the place you want to submit). Then, we'll have a conversation with you about whether using GenAI to complete your project is the most effective way forward, how you can use it to support your learning and writing process, and how to use it without violating any policies. Because consultations are confidential, we will not report to your professor or other audiences that you used GenAI.
• Will the CWE use GenAI while giving feedback?
  • No, the CWE will never use GenAI to generate feedback, especially not in written consultations. We believe that no work should be submitted to an LLM without the consent of the writer. Although some consultants may use GenAI as a tool during sessions—for instance, brainstorming ideas—feedback for writers will never be generated by GenAI.

For faculty

• What happens if one of my students brings work to the CWE that used GenAI?
  • Our consultations are confidential, so we will not report to you whether your student is using GenAI. We advise students to abide by the Academic Honesty Policy and whatever course policies you have set out regarding use of GenAI. If they are allowed to use GenAI, we will also discuss with them whether using GenAI is the most effective route for completing the assignment and improving their skills. All of our staff receive on how to discuss GenAI in consultations. If we have concerns about the writer’s use of GenAI, we direct them to discuss it with you.

• Does the CWE use GenAI detection software?
  • The CWE does not use GenAI detection software. We aim to humanize instruction and always assume writers are presenting their own work unless they tell us otherwise. If we suspect a student may have misused GenAI or violated the Academic Honesty Policy in some way, we always point out our concerns, provide pathways for revising their work to align with university and professor policies, and encourage them to discuss their concerns and use of technology with their professor.

At this stage, many studies are finding that existing detectors are not always accurate at detecting use of GenAI and generate many false positives (see, for example, these recent studies by , , and ). also gives a nice overview of some of the concerns and mentions the stance of The Modern Language Association/Conference on College Composition and Communication (MLA/CCCs) and the Association for Writing Across the Curriculum (AWAC) in their respective statements on AI. Therefore, we recommend that faculty never rely on detectors alone when they suspect a student may have misused this technology.>

• If detection software is unreliable on its own, what else can I do?
  • All that being said, the question is: so now what? How do we address suspicions of AI use without falsely accusing students or damaging the professor-student relationship? In the CWE, we've been advising faculty to approach it like other suspected plagiarism cases where there is no clear proof. Some possible steps are below:
• Meet with the student and ask them to discuss the paper's content with you. If they struggle, ask them if they used AI or other resources that they did not cite.
• Meet with the student and explain your concerns, giving them the opportunity to disclose AI use.
• Depending on students' responses to items #1 and/or #2 above, decide next steps. If they insist they did not use AI but struggle to articulate the ideas in the paper, you could still assign a lower grade or zero and ask them to rewrite the assignment. If they do admit use, you could assign a zero or lower grade, and if you're willing, give them a chance to rewrite.

When meeting with students, it can be helpful to have some evidence ready to discuss, even if it's not evidence that you can prove directly as with other kinds of plagiarism, such as when a student copies and pastes from another source. Some evidence could be:

• Running the assignment or topic through ChatGPT a few times and saving the responses to show similarity or overlap with the student's submitted version.
• Comparing the student's assignment with earlier writing samples and identifying discrepancies in vocabulary, style, tone, etc.
• Identifying common ChatGPT phrases/words/organizational structures and showing the student comps generated by AI.