Professors blindsided by new ASU AI tool that chops up their lectures and uses them out of context

Atomic is a new Arizona State University platform that creates AI-generated versions of ASU faculty’s lectures, cutting down often long videos to short clips for quicker learning. The idea is to give students personalized learning modules tailored to their schedule.

But Sam Cole reports many ASU professors are disturbed by the way their lectures are being chopped up and used out of context — and they were blindsided by the platform’s launch.

Cole is co-founder of 404 Media, where she covers the internet and AI and she joined The Show to talk about it.

Full conversation

LAUREN GILGER: Good morning, Sam.

SAM COLE: Good morning. Thanks for having me.

LAUREN GILGER: Thanks for coming on. So tell us how this is supposed to work. It’s in beta now, we should say.

SAM COLE: Yeah, so just from the the little bit of testing that I’ve done and hearing from faculty that have tested this out also, what you do is you kind of prompt like you would ChatGPT. So you open up the ASU Atomic platform, it says "What do you want to learn about today?" and in my case I said "I want to learn about ethics in AI." So I typed in ethics in AI and it asked me how much time do you have to learn about this, what’s your level, what do you want to learn, how quickly, things like that. And then it creates it generates a customized, "learning module" for each person who’s using the platform. And then from there you get sections and readings and videos and all that stuff.

LAUREN GILGER: OK. So the the videos in particular are coming from, it sounds like, longer versions of ASU professors' lectures on on various topics?

SAM COLE: Yeah, so as far as I can tell, it’s coming from Canvas videos. So Canvas is obviously the learning management system that ASU uses. And professors upload lectures to Canvas that students can then watch later or during class, whatever case it may be. And the lectures are long and in context of the class. So you know, you might take like an hourlong lecture that’s a very complex topic. Obviously in my case I was testing ethics, that’s a very complex issue in general, especially in AI. So it’s taking those really long lectures and then turning it into 40-second to 2-minute snippets. Where I assume the platform is, you know, finding, "Oh, this professor mentioned the definition of AI." Even though maybe the class was about film, the professor mentioned AI, so they snippet out 40 seconds where the professor defines AI and then plug it into this learning module. And then from there it generates text based on what that snippet said. So it’s taking complicated things —

Sharon Attia

/

Handout

LAUREN GILGER: Right. And that’s an actual example that you found in the module that it created for you? Like a film class and a like a 30-second or 40-second clip about AI?

SAM COLE: Yeah, exactly.

LAUREN GILGER: OK. So tell us about what the professors that you spoke with had to say. It sounds like this started for you with a post someone posted on Bluesky?

SAM COLE: Yes, so Chris Hanlon, who is a U.S. literature professor at ASU, posted about this on BlueSky and I think on Facebook talking about how ASU had rolled out this new thing called Atomic and that the faculty that he had spoken to had not heard of this before, he had never heard of it.

He logged on to it and was like "Oh, this is my face, these are my lectures, these are my colleagues' lectures." As far as I can tell, most faculty hadn’t heard about it or weren’t aware of it being developed before Friday, so a couple days ago.

And one one scholar that I talked to said, you know, "I hadn’t heard about this until you emailed me about it." And then I was walking through the phone on the phone with them saying here’s how your lecture was used in this AI program and they were horrified. So most people just hadn’t heard about it until professor Hanlon had mentioned it online, but then especially um in the last few days where it kind of gained some more traction.

LAUREN GILGER: So some tension around the rollout for some of these professors you spoke with. Are they also concerned about sort of their intellectual material, their their original ideas? I mean do they own those to begin with I guess if they’re posted on Canvas?

SAM COLE: So it’s tough because when you work at a university a lot of your work is owned by that university. And you know, the university might have intellectual property rights to your work, but it doesn’t have exclusive rights in a lot of cases. So even if ASU does own — and as far as I can tell it does own the content that goes up on Canvas, and if you’re employed by ASU obviously they have a right to that content — but what was concerning was the the total lack of context and credit and further reading even, ways for learners to follow up on that content.

So in a way it’s like they they own — ASU might own that content, but it’s stripping the ownership of the thoughts and the preparation that went into that content from the people who created it, which would be the professors and the faculty. So it is an ownership issue, but it’s also the the fact that the faculty that I talked to were saying, you know, "I’m not trying to be compensated for this, although that would be amazing, because professors are not paid enough in general across the country. But, you know, it’s not about credit and compensation, it’s about reflecting the thought and the time that I put into creating these materials for my students because I care about them, then you know stripped and cannibalized and chopped up into slop for AI" is very concerning for these scholars. So I think that’s that’s the really the issue for a lot of folks.

LAUREN GILGER: This comes, Sam, as AI is really kind of upending education in a lot of ways. Like whether it’s at the university level or or down the line, but I think ASU in general has been really positive about AI and its potential in education. It sounds like that’s what the goal is of creating a platform like this. Tell us about that kind of context in which this is happening.

SAM COLE: Yeah, so I mean, AI is obviously something that’s being pushed into or embraced by — however you want to look at it — institutions and organizations and industries regardless. It’s academia’s dealing with this, you know, there’re tons of different industries, my industry, journalism, is is dealing with how to grappling with how to use AI and if to use AI. So I think that’s the context that ASU is trying to keep up with, they’re saying, you know, "We don’t want to be behind on AI innovation and AI-powered tools. And we want to be able to offer these things" to what is essentially a public product.

It’s like I was signing up for it, it’s $5 a month after a free trial, it’s not for students or ASU community, it’s for the public. So they’re trying to stay ahead so to speak for for the sake of staying ahead. But it’s not actually something that’s providing value in a way that reflects ASU’s actual quality education.

It’s definitely what’s being produced is not not good honestly, it’s pretty bad content and full of errors. So even if it was like a good product, I think it would be questionable. But the fact that it’s a bad product and rolled out pretty, you know, halfway is just even more concerning.

LAUREN GILGER: All right. Well we’ll leave it there for now. Sam Cole, co-founder of 404 Media joining us to talk about this story. Sam, thank you for coming on, appreciate it.

SAM COLE: Thank you so much.

LAUREN GILGER: And KJZZ reached out to ASU for comment, received this response: "A pilot launch of the program began in April, the pilot explores how ASU can use existing digital content in new ways to reach learners beyond those enrolled in degree programs."