[Courses] 3 very easy things and one difficult thing Canvas can do to help prevent AI/ChatGPT abuse.
ChatGPT and other AI language models threaten online education and by extension the usefulness of Canvas. It is in the interests of educators and infrastructure to take this threat seriously, and, while there is no silver bullet, there are some very simple things Canvas can do to help instructors preserve academic honesty. Example case: Students are assigned a discussion question, cut the instructions from Canvas, paste it into ChatGPT, cut ChatGPT's output, and then paste it into the forum. There are three very easy things Canvas can do practically over night that would make this kind of academic dishonesty more difficult for students.
These are the ideas in order from easiest to implement to the most difficult: Very easy to implement: 1. Allow the instructors to prevent students from cutting text from assignments. Make it so that this is prevented only from the student side when selected. How it helps solve the problem: This will prevent students from easily cutting assignment directions and pasting them into ChatGPT and related programs. This is simple to implement. Many websites block the ability to cut and paste. Yes, students can type up the instructions themselves--but students who cheat are lazy, and less likely to do this. 2. Allow instructors to prevent pasting into assignments (again only from the student side). How it helps to solve the problem: Similar to the previous suggestion, this makes it more difficult to just copy material over from ChatGPT. Again, students can type all the material manually, but they are less likely to do this since they are the students who are lazy enough to cheat in the first place. 3. In line with the previous options, have the ability to simply TRACK if the material was cut or pasted from particular pages, even if pasting is allowed for students. How it helps: This serves as additional evidence of academic dishonesty if there are already other reasons to suspect academic dishonesty. More difficult to implement: 4. Allow instructors to set up autoscan on discussions and all written answers in tests and quizzes for use of AI. Preferably, make it so that instructors can select the service that does this autoscanning. This is more difficult to implement, but the first three suggestions are dead simple. They could practically be implemented overnight, and would be VERY effective for helping curb academic dishonesty. We need this much more than we need cosmetic changes or threading in forums, or many of the other priorities infrastructure seems to have been working on. AI learning language models are threatening online eduction--and that in turn threatens the value of LMS systems like Canvas. It is in everybody's interest to implement these low-cost and low-effort changes that will help deter such behavior.
instructor
