Rethinking Academic Honesty in the AI Environment
One of the biggest challenges we are facing as instructors now is ensuring that the content submitted by our students is an authentic reflection of their work and learning. As shown on the following page, there are ways we can augment our teaching approach to better ensure authentic assessment, but unfortunately many of us will find ourselves in situations in which we are questioning the authenticity of our student's work.
It is important to be very clear with your students about what uses of generative AI are allowed or not allowed in your course through a clear AI policy in your syllabus. But if you believe that a student may have broken the rules you set forth, there are a few tools that you can use to help guide your discussion. Turnitin has some helpful advice for having these tough conversations Links to an external site. with students as well as some advice for students to ensure that they are using AI ethically Links to an external site.. This helpful Guide for Suspected Academic Misconduct from McMaster University Links to an external site. may also help you in this process. Quill.org's toolkit Links to an external site. has a clear conversational template.
Detecting Use of Generative AI in Assignments
There are some signs that the work submitted may not be a student's own. You will become much better at spotting these as you become more familiar with generative AI tools yourself. Generic phrasing, bullet points, and odd formatting can be good tip-offs. Our colleague at LAVC developed this handy Faculty Cheat Sheet for Spotting AI Writing (J. Dillon, LAVC). Download Faculty Cheat Sheet for Spotting AI Writing (J. Dillon, LAVC).
There are several AI Detectors that are now available online in which you can copy/paste text and get an assessment of whether it is more likely human or machine generated. However, these come with a big warning: none of the AI detectors that exist today are 100% accurate. Each of these tools will provide you with a different determination Links to an external site. of the probability of a submission being AI generated. Even then, it is a probability and not a certainty. False positives happen, and they may happen more frequently with students who are English language learners Links to an external site..
The images below show two different AI detectors that have analyzed the exact same AI generated text: one reports 0% and the other reports 100%.
As new AI programs are developed, these tools become more and more obsolete, so in the long run, we will be able to be more effective instructors by revising our teaching and incorporating more authentic assessments (see the next page for more!). That way, we can spend more time ensuring that our students are learning and less time policing them. Still, if something seems off about a submission, here are some tools to help give you information to guide your conversation with your student.
Within Canvas
- Turnitin has a Generative AI detector Links to an external site.. When you use Turnitin for your Canvas assignments, you will be able to see an AI probability score. Again, this is not 100% accurate and should always be a prompt for a discussion rather than proof of dishonesty.
- Harmonize: Links to an external site. this Canvas plug-in enables you to create discussions that integrate with TurnItIn's plagiarism/AI detection.