Remove the AI Summarize button from Discussions as soon as possible.

[william_simkule]

I noticed the Summary button yesterday, and thankfully found out this was not available to students.

This "feature" needs to be removed.

Generative AI might have it's uses, but there is no reason for a student or instructor to use generative AI in a discussion forum.

Generative AI without citation is plagiarism, and proper citation probably includes a complete list of works and sources the AI has been trained on.

Yesterday I used the [Summarize] button on a discussion with only one post, the sample post.

The assignment instructions ask the student to select a luxury that Peter Singer's weak principle from Famine, Affluence, and Morality would require they give up to prevent "very bad things."  My sample post does this, entertaining the idea that I give up going to the movies in exchange for purchasing a life-saving drug.

The AI generated Summary, though, talks about Peter Singer's Utilitarianism and (falsely) claims that I've applied it to the topic at hand.  The term "utilitarianism" does not appear in the post or instructions; rather it is the theory Singer is well known for advocating... just not in the assigned readings.  This is to say that the generative AI probably searched the internet for information about Peter Singer and injected it into the summary.

There may be some uses for generative AI, but grading isn't one, and apparently summarizing isn't one if it's going to search the internet, rather than the discussion posts themselves.

I understand that someone probably gave you a lot of money to waste on generative AI.  If you can find a use for it; great!  But this isn't that use.

[fplunkett]

I see your point. I have used this summarize feature in several discussions in the last weeks of my last session. I did not notice anything that was brought in from outside of the discussion as my students' posts were very thorough. that said, maybe there needs to be a control to be sure that it ONLY summarizes the posts in the thread. What I did was create the summary, and then copy and paste it into a new reply. I then edit the reply to my liking before posting. My students only see what I choose to post. I think it is a helpful tool for me. I would, of course, remove anything that would constitute plagiarism before posting. 

Frank P. 

[william_simkule]

To be clear, this is a "summarize" feature in name only.

If I'm reading you right; you didn't actually bother to read your student's posts, but rather had an AI tool "summarize" them for you and then you "edited it" to your liking.

But let's be clear.  You plagiarized an AI writing tool w/o giving it credit, so you're setting a bad example for your students.

Plagiarism occurs when you present another's work as your own, and here you're presenting the result of the programmer's work as your own.

If I was a student and I found out that you graded my work without even bothering to read it, I think my immediate goal would be to make sure that you're removed from the course in favor of a replacement that will bother to read my work.

[fplunkett]

Thank you for your comments. I may have not been "clear" enough. I did read all of my students comments before using the summarize feature. I was confident that the AI generated "summary" a summary of my students posts and nothing more. I edited to add anything I thought needed to be covered more thoroughly. I guess it is debatable as to whether I "plagiarized" by using this tool to summarize the discussion thread. To your point, in preparation for an upcoming professional development session on AI, I have noted and will share your concerns about the content that was apparently brought in from the outside. I will be counseling my fellow teachers to be diligent in inspecting the summary should they choose to use it and be sure that all of the content is only a summary of the discussion. I will also make a note to post a reference to the Canvas AI summary in future uses of the tool. 

Thanks again. I appreciate your perspective.

Frank Plunkett

[jbielecki]

@william_simkule, I, too, don't want or need AI to "summarize" my students' posts (or whatever it's doing). It is my job as their instructor to read their work. I had no idea this feature was coming and couldn't find anything about it when I searched the Community for "Open summary" (in quotation marks). After digging through this Community "haystack," I found this: https://community.canvaslms.com/t5/Discussion-Summaries-Feature/Discussion-Summaries-FAQ/ta-p/609172. I hope it helps you as it has me.

[bec_plumbe]

William, this is very troubling. I'll do my own testing of course, but it would be very helpful if we could start compiling examples here. If you are able to share screenshots without privacy / copyright concerns, that would be instructive.

As an educational designer, I have wondered about leaving this feature on for teachers. Most of them will stick with the default settings of the site as they find it. This makes me think it is probably preferable to turn it off as default, leaving the individual teacher to turn it on if they seek it out.

[adamwarecx · Accepted solution]

Hi Everyone,

I'd like to take the opportunity to direct you to the FAQs (https://community.canvaslms.com/t5/Discussion-Summaries-Feature/Discussion-Summaries-FAQ/ta-p/609172... especially the AI Nutrition Facts (https://community.canvaslms.com/t5/Artificial-Intelligence-in/AI-Nutrition-Facts-Canvas-Discussion-S...) which provides information on where and how the AI is being used in the functionality

 @william_simkule - This is a per instance Feature Preview (Feature Previews are feature options in active development.) and therefore be turned on/off by admins (account level) or by teacher (course level). I'd suggest discussing it's use for your students, subject, department and school within you institution first. It's also worth noting that this functionality is not available for students as per the FAQ.

We appreciate all feedback.

[kyh802]

I was completely appalled to find this tool. Thank you for highlighting how incredibly inappropriate it is in the context of a learning management system.