Academic Integrity & AI
Near the top of the list of most educators’ questions about AI is, "How can I prevent students from using AI to cheat?"
To address this question, let’s break it down into three relevant questions:
- Can I create an AI-proof assignment?’
- Can AI-generated writing be detected?
- and, What should I do if I suspect a student is misusing AI?
Unfortunately, just like many academic integrity questions, there are no simple and concise answers. Instead, we must reflect on our individual expectations, context, and moral beliefs to ensure a well-defined policy for AI and academic integrity.
While creating an AI-proof assignment may not be feasible, AI tools' rapid and impressive growth presents a projection of near-limitless potential. Our goal is not to create a barrier to learning but to encourage students to learn with and without the help of AI tools. Therefore, we recommend creating authentic assessments, promoting process over product, and establishing guidelines that discourage accidental AI misuse.
- Additional guidance for developing process-focused assessments can be found in the AI Assessment section of the Center for the Enhancement of Teaching & Learning's Bearcats Landing site (UC login may be required).
- TurnItIn provides a useful AI misuse checklist to help prevent AI misuse.
- AI misuse rubric can help guide designing AI-Aware assignments.
The research suggests this answer is still up for debate. Some research has concluded that it is impossible to accurately detect AI-generated writing, resulting in OpenAI abandoning its own authenticator, saying it does not believe it is possible. Other research has highlighted the impressive results of some academic AI checkers, most notably TurnItIn’s AI detection tool. Although there is uncertainty, we can draw two helpful conclusions from the growing body of research:
- AI detection tools designed for specific populations and types of writing can be performed with high accuracy.
- Even if we cannot be 100% certain of AI-generated text, we should approach potential academic integrity violations with the same care and through the same procedures as previously.
If you suspect a student submitted AI-generated work without permission, it's important to follow the same procedures you would with other academic integrity violations. However, given the relative newness of these tools and the ongoing confusion and disagreement on their proper use, it's suggested to approach potential AI misuse with more intentionality and grace, fostering a culture of understanding and empathy.
- Carefully Analyze the student work – what other issues become apparent, commonly the result of AI. If using TurnItIn, familiarize yourself with their AI detector with this guide.
- If AI misuse is suspected, it is highly recommended that you reach out to the student. It is prudent not to accuse the student of AI misuse, but to inquire about the irregularities in the assignment. Here is a useful guide on how to approach students regarding potential AI misuse.
- If you feel that the infraction violates UC Conduct Policy, treat the academic misconduct as any other violation and follow college policies for reporting misconduct.