2 New Ways We’re Using AI to Streamline Workflows in Canvas

zachp
Instructure
Instructure
5
2824

Instructure.png

Together with our customers, we've identified several areas where AI can profoundly enhance our platform. One of these is using AI to build efficiencies and increase capacity for educators and administrators. Specifically, AI can streamline tedious tasks, allowing educators to focus more on engaging directly with students. With this in mind, I'm excited to introduce two new features within Canvas (currently in limited beta) designed to save educators time.

 

Feature Spotlight: Discussion Summaries

discussion-summaries-shortened_2.gif

The Discussions feature in Canvas is a powerful tool to facilitate conversation and collaboration between students. But we know it can be challenging for educators to keep up with threaded discussions, especially those with larger classes. That's where discussion summaries can help – they accelerate educators' ability to grasp central themes and questions emerging from discussion threads, understand student engagement, and identify areas that warrant further conversation. This feature is currently available as a limited beta, with plans to release more broadly in 2024. 

 

Feature Spotlight: Writing and Editing Tools in the Rich Content Editor

rce-ai-tools2.gif

 

Educators and instructional designers spend hours creating course pages and materials that cover their desired learning objectives and meet the unique needs of their students. To speed this process up, we're testing new AI-powered writing tools within the Rich Content Editor. These tools can help summarize text, identify and define key terms & phrases, change tone or text complexity, and more. This feature is currently available as a limited beta, with plans to release more broadly in 2024. 

 

Stay tuned for another AI update from me next month leading up to some big reveals at InstructureCon 2024 in Las Vegas

5 Comments
mmoore1
Community Contributor

"To speed this process up, we're testing new AI-powered writing tools within the Rich Content Editor. These tools can help summarize text, identify and define key terms & phrases, change tone or text complexity, and more. This feature is currently available as a limited beta, with plans to release more broadly in 2024." 

So is the currently in beta?  I have access to that if it is.  

zachpendleton
Community Participant

@mmoore1 — unfortunately it isn't in everybody's beta environment quite yet; it's being used by a small group of institutions who are providing us feedback and helping us to refine the feature before we release it more widely. Stay tuned for details!

mwolfenstein
Community Participant

I want to say this early and I will say it often. AI summary needs to be an optional tool in any course for foreseeable future. I know there's a huge push to have Gen AI in products right now to seem current and cool, but we're messing with a technology where we really don't understand what the implications are in relation to learning.

There are two things that I can confidently assert:

  1. When I've used LLM AI to generated summary content in my own work, it misses huge important things. It's still a net win for my workflow because I know the primary source material so it saves me time, but it would be disastrous for me if I had to hand just the AI summary to someone and assume that they took away all the key points. If you consider the implications for both faculty and students, this is deeply problematic. Faculty relying on these tools might miss really important contributions from individual students, and students relying on them might miss foundational understandings or essential elements.
  2. When faculty are assigning students to engage deeply in a discussion forum, the point is often times to engage, synthesize, analyze, and generally develop critical thinking skills. When we use an LLM AI to do that analysis, we're not learning. Using these tools when we're already competent in terms of the underlying skills can be awesome, but using them in place of developing those skills is an incredibly dangerous path for us to go down.

Personally, I think that this is not the best use of AI in the Canvas platform. There are undoubtedly many instances where Gen AI can be a powerful tool for teaching and learning with Canvas, but I advise the utmost caution when it comes to the use case of summarizing content.

 

zachpendleton
Community Participant

thank you for the feedback, @mwolfenstein! we’re committed to this and all our AI features being disabled by default to allow institutions and educators the choice to what’s best for their classrooms. 

as regards this feature in particular, two things: (1) it’s been designed as instructor-only because we agree with your points regarding use by students; and (2) I think the most powerful use of this feature is in large format classes where discussions were difficult to use due to volume of responses, and our hope is that an instructor summary will allow the educator to use discussions and understand student opinions and insights in places that weren’t previously feasible — not that it will become a substitute for engaging directly with students. 

thanks again for the feedback, and I hope this assuages some of your concerns. 

mwolfenstein
Community Participant

@zachpendleton This is definitely good to hear and greatly assuages my concerns about where this AI tool is going. I'll still assert that I'm seeing a whole lot of rushing going on in basically every sector to implement Gen AI solutions, and personally I think that there's a lot of moving too fast and a lot of things are getting broken without consideration of the consequences. The potential for advanced AI as a teaching and learning tool is huge, but I hope that we can all take a measured approach to the greatest extent possible.

For Instructure, I understand the need to play in this space and show that you're working towards adoption and integration of AI solutions. I'm seeing some general improvements in product from you all in general, but I hope that the rush to AI doesn't detract from addressing other long standing issues like the array of extant accessibility issues, lack of parity on mobile app features where parity is viable, and finishing development of incomplete features like the assignment comments library.