Colleges are experimenting with making online teaching videos featuring AI avatar versions of professors. Some students find the simulated likenesses of their instructors a bit creepy, but proponents say the technology could be key to making college courses more active and human. The idea is that AI will make it easy to make personalized teaching videos so that more teachers can adopt a “flipped classroom” approach — where students watch video lecturers as homework so class time is spent on discussion or projects.
Demo of Sara Cochran’s AI avatar, University of Virginia website
Sora and Vibes: AI Video Now Officially Going After Social Media, Platformer
Jeff Young:
On the first episode of this podcast, I said something that turned out to be controversial, and it led a couple listeners to email me challenging what I said.
Rob MacLeod:
My main pushback comes in response to your hypothesis that we professors only want to use Gen AI for the things we like least about teaching.
Jeff Young:
That's Rob MacCloud. He's a professor of biomedical engineering and internal medicine at the University of Utah. On that first episode, my guest was an expert on learning and AI, Mutlu Cukurova, who talked about how worried he is about careless uses of AI in education that don't take into account how different AI works compared to how human brains work. And in the interview, I was surprised when the professor said he would be totally fine with an AI avatar standing in for him in a lecture video, because he said making videos is tedious. He talked about the power of synthetic media, where AI could make a video version of him that looked strikingly realistic.
Mutlu Cukurova:
We’ve been experimenting with AI generated synthetic media, looking at if the same content is delivered by me recording my video in front of a screen, or ex human expert recording the video, versus the same content delivered with AI generated synthetic media. And in our experimental results, we have shown that there is no statistically significant difference in terms of the learning gains.
Jeff Young:
To me, this seemed like a teacher handing over a key part of his job to a bot just because he didn't like that part of teaching and wanted to focus on other things. And so I made this big argument that it seems like everyone is okay with automating some part of their job to AI, but that no one can agree on which part is okay to hand off to a bot and which parts are key and should definitely only be done by humans.
But Rob, the listener who wrote in, argued that I was missing something important here.
Rob MacLeod:
First, I think you might have too limited experience with flipped learning approaches, which I've used extensively since the pandemic. I was trained in sound engineering, and so it was an easy and pleasurable move to make videos in a little studio I set up in my home. However, I also have to acknowledge that the time it takes to make, edit, update, etc, these videos is a serious obstacle in convincing my colleagues to consider this format — a format they usually agree can be more effective. This overhead also gets in the way of my own revision of that content.
Jeff Young:
As you probably know, flipped learning is this teaching strategy where professors record a short video lecture with some key background information, and then assign that video as homework for their students, so that class time can be spent on more active things like group projects or discussion instead of doing lectures in class.
Rob MacLeod:
I completely disagree with your statement that standing in front of a classroom imparting hard-won expertise at what makes us professors. We know that this is the worst way to teach. You certainly know this too, and have likely seen the learning pyramids that place typical lectures at the top. They're like sugar in the food pyramid. Things we should consume minimal quantities of lecturing is not what makes good professors. At least not what I have discovered is my main and most effective purpose.
Jeff Young:
So this email by Rob ended up sending me on a journey to learn more about this idea of using AI to quickly spin up teaching videos with an avatar version of the professor. Is this going to end up being one of the big ways AI comes to education?
Hello, and welcome to Learning Curve, a look at what it means to teach and learn in the age of generative AI.
I'm Jeff Young, a long-time journalist exploring the intersection of education and technology.
So when Rob reached out, I was super curious to hear more, and it turns out he is someone who has spent a lot of time over his career as a professor trying to improve his teaching and to make it work better for his students. A love of teaching is what drew him to being a professor in the first place. But early in his time working at universities, he realized that research was the activity that was going to be most rewarded.
Rob MacLeod:
The challenge we face, the stress we're under, is trying to juggle that desire to do a good job teaching with the realities of the need to deal with all the other parts of our job.
Jeff Young:
And when we chatted, he reminded me that professors do more to support teaching than many people might realize.
Rob MacLeod:
We, you know, we are on the committees. We're coming up with curriculum. We're dealing with all aspects of teaching. You know, I'm involved. I've been the director of an undergrad program for almost 20 years. I deal with, you know, stuff like, how do we set up the schedule next semester? And and. And which electives are approved, you know, for the students. So there's, there's many, many hats we all have to wear in doing this job and teaching. You know, sadly, is the one where we're, you know, we're sometimes forced to truly cut corners, or at least to be as efficient as we can be.
And it's that trade-off between trying to be effective and efficient at the same time that I think, you know, AI has at least some potential to help us with
Jeff Young:
During the pandemic, when so much was shut down and classes moved online, Rob says he finally had the time and opportunity to try a flip teaching approach that he had been reading about and wanting to do for a while, and as things went back to in person teaching, he had to get creative to find a space where he could break his class into groups for discussions and hands on work that wasn't in a traditional lab.
Rob MacLeod:
Well, I created a classroom. I found a room that was not as set aside as the classroom space. It was kind of an undefined lab space in our engineering building, and and my colleague tipped me off that, oh, there's this room, and, you know, might work for kind of small teaching like you're talking about. So I found this room. I glommed on to it.
I managed to get it sort of officially designated to our department. And then I went over to we have a like all universities, we have a facility that collects old furniture and things that need to be you know, that are available for reuse on the campus right, that you can get for free.
So I went over there with my again, my colleague helped me with this. We found some tables, we found some chairs. I bought a screen and a projector and a few white boards and put them around the room.
So I created this space, and I said, this is enough space for four tables, eight people per table, 32 students. I'm putting 32 chairs in here. No more exactly, and they fit. Everybody can fit in that space and and that's how we're going to do this. And so that way, no matter how big my class gets, we still work in sections that are manageable, because I don't think you can do what I do with 100 students or even 50 students in a room. You need a small group to multiple small groups to actually make this work.
Jeff Young:
Since then, he has created more than 200 videos for his students to support his flipped teaching experiments, but he sees that many of his colleagues, they might not have as much knowledge of video and audio production as he happened to have,
Rob MacLeod:
But I want to make it easier than I had to go through to facilitate this form and so. So when someone comes along and says, Oh, here's the way to make those videos, or here's the way to extract content and turn them into reasonable videos in a reasonable timeframe. That, to me, is a bonus that that gives me an incredibly valuable tool that I can go to my colleagues with and say, Hey, this part of the job suddenly become a lot easier because, because, you know, everybody gets very intimidated, even technical people with, you know, once you make one, you know, once you make one video and you edit it and you do all that work. Oh, my God, you can't scale that.
Jeff Young:
So he hopes that AI might cause a big win for flipped teaching, which he says really has worked for his students. And he wrote in not just because I was dismissing this idea of these video avatars, but he objected to my implication that professors are all looking to somehow find shortcuts just to make their own lives easier,
Rob MacLeod:
And that we think more deeply than what are the parts of this job I don't like, so I'll get AI to do that. I think we're getting to the point as we learn, and this is early days, but as we learn, we're getting to the point of saying, Oh, here's a part that AI could help me be more efficient with that would help me save time for the things that I think I'm truly effective at doing. And there's lots and lots of research data that, as I say, there is a pyramid of teaching effectiveness, and the best is peer to peer education, and the next best is people doing things. People actually building things. And I teach biomedical engineers. So they're engineers. They want to do things, they want to make things, they want to they want to solve problems. So the more guidance and time and space I can provide for them to do that at the cost of me imparting whatever wisdom I might have. You know that's that's so much better time spent.
Jeff Young:
It's turns out that Rob wasn't the only listener who was put off by my deep skepticism of the idea of Avatar teaching materials. Soon after Rob emailed, I got another listener email, this time from someone I know pretty well, and who I've talked to many times over the years for my reporting. So one note I had on your earlier episode, that's Derek Bruff, Associate Director of the Center for Teaching Excellence at the University of Virginia,
Derek Bruff:
The use of AI generated avatars is something that some of my UVA colleagues are actually big on their argument. Is that the professor focuses on writing the script, and the AI does all the video work thereafter. And so if you need to change the script, that's a lot easier than re shooting a traditional video.
Jeff Young:
Derek pointed me to one of those colleagues who are actually making AI avatars, Sarah Cochran. She's the senior director of learning experience and Digital Innovation at the University. Before I got to talk to Sarah, I first met her AI avatar, which is featured on a UVA website, demonstrating their experiments with the technology. The demo is essentially an AI chat bot interviewing her AI avatar about her professional background.
Sarah Cochran:
My background in music is a bit niche, and like a lot of folks in higher ed, I thought I was going to be a professor, but after my TA funding ran out at the University at Buffalo, I took a job as an academic advisor at Southern New Hampshire University. That's where I discovered instructional design and realized it brought together everything I love. If
Jeff Young:
You watch this avatar and weren't paying close attention, you might think it's the real college administrator. There are some few minor oddities, though, in how the avatar moves, like one time when her head moves forward for no real reason between sentences. In some moments like that, it felt a little creepy. In what has been described as the uncanny valley, when virtual technologies are realistic but not quite perfect, and it can trigger unsettling feelings. When I set up an interview with the real Sarah Cochran by zoom, I started by asking what motivated her to get into making an AI version of herself.
Sarah Cochran:
It's so it can be so expensive and costly to produce a video that doesn't look like you recorded it on your, you know, right around your phone and stuff like that. So and you have to learn the the technical like upskilling, or the knowledge that you have to gain to even be able to do some sort of post production on it is the learning curve can be steep depending on what tool you want, you want to use, right? And so it's, how can we make it cheaper, still produce a decent product, while also kind of like democratizing the access too. So if I'm a faculty member that maybe isn't technologically savvy, that's absolutely okay. How can I create a video that looks pretty, not pretty good for my class, that's easy to edit, and I don't, you know, I don't necessarily need help from anyone else to do that. She
Jeff Young:
has worked on lots of old fashioned video shoots for online courses she's helped make at her previous roles at the Southern, New Hampshire University and northeastern including one course that she says cost well over six figures to produce, though, she noted that not every course was that expensive.
Sarah Cochran:
Another huge benefit of these tools is the dubbing, so that way, if you wanted to translate it into different languages in with ease. That's another huge benefit. So if you were to create, you know, your video in Synthesia or Hagen, and then you were like, I want to offer a class or a section of my class to you know, people who are native Spanish speakers, you can then dub all of your videos into Spanish, and it would actually align your your lips with the dubbing as well.
Jeff Young:
Okay, I have no idea what you all are thinking out there, but I had so many questions about what it's like to make one of these avatars and how students react.
Sarah Cochran:
I have a student intern right now, and so I made him watch a lot of our avatar videos, and I was like, How do you feel about this? And he's like, I don't know if I'd want an avatar of my professor.
Jeff Young:
And why? What do you get a sense of? Why? Just because it's too is it that uncanny valley, like it's a little eerie?
Sarah Cochran:
I think it's about the feeling that maybe the professor didn't want to do it.
Jeff Young:
What's the word, phoning it in.
Sarah Cochran:
Now, I don't know about phoning it in, but it's like you're losing that. I think maybe it is the uncanny valley, because you're, you're losing that connection to your professor, because they're there, but they're not, right, right, right. Yeah, they did create that video. Or did they, you know, or did someone else help them to create that video? So now all you have all these questions as like a student in your head, especially if it's like wearing the face of your professor.
Sarah Cochran:
Okay, I have something I would love to do, if you're game, which is, I'm gonna, do you mind if we watch together the short video that you have?
Sarah Cochran:
I’m really excited for this conversation. My background in music is a bit niche, and like a lot of folks in higher ed, I thought I was going to be a professor, but after my TA funding ran out at the University at Buffalo, I took a job. Okay, so I'm
Sarah Cochran:
going to stop it, because not to watch the whole thing. But when you see this, I mean, I don't even know if you were even looking. It's like, do you, do you have a sense of like, Oh, can't even look. Or do you, what do you think when you see. Yourself, but it's not you, you know, you have this kind of spun up from, you know, it's a likeness of you that, like you said, is AI generated based on your script as I …
I mean, like, if we're gonna be completely honest, I Hate it. Hate Everything about it. But mostly it's because, like, when you listen to your own voice, you don't really like listening to your own voice. So then when you not only hear your voice coming out of, you know, out of just listening to your own voice,
Jeff Young:
Just to be clear, the voice you are hearing in this AI avatar is also an AI simulation. All Sarah had to do to make it was to type a script into the system,
Sarah Cochran:
And then, and then, on top of that, it's your face, but not necessarily moving the way that you would normally move in the situation. It's, it's, it's very strange.
Jeff Young:
As I was watching it, I was kind of looking like, I mean, it's a pretty good, pretty good likeness. It's like, pretty it's tracking really well. It doesn't feel like it. I guess if I really didn't know, I'd be curious if listeners watch it what they think. But if I really didn't know what it was beforehand, it could pass. Probably it'd be believable, at least in a short clip.
Sarah Cochran:
I sent it to some of my friends after I made it, because I was like, I don't know if I should include that with the rest of the interview or not. And and they're like, Oh, I didn't know that wasn't you as first. And I'm like, you know me. How did you not know that was me?
Jeff Young:
She says that with the AI tools out today, making an AI avatar is pretty easy.
Sarah Cochran:
We've compared different tools. We have subscriptions to a bunch of them, just to see like strengths of of each one. What can, what can they do? Well, what can they, you know, not so don't do so great at and then we started using them and, you know, kind of piloting or proof of concepting them in various situations with partners around. Are you frozen? No, oh, sorry, wow. Just sitting still, standing still.
Jeff Young:
Okay. This was a strange moment in this conversation. I was there on Zoom, listening carefully, but I guess I was so still that Sarah thought the system had frozen and that I wasn't there anymore hearing her. I was just reminded I sometimes tap on my desk and mess up the microphone. So I was trying to sit still, so
Sarah Cochran:
still, and I'm like, Gosh darn it, my internet
Jeff Young:
Anyway, we shook that off and got back to talking about how to make these avatar videos.
Sarah Cochran:
And honestly, the quality could be even better. I recorded that in my at my house, on my computers. You recorded the audio. You mean, no, so, like, basically what you have to do, this was created using Synthesia. So what I had to do is read a script that Synthesia provided, and you're supposed to, like, do the different the words, and then it tells you make sure while you're reading it in this section, you're angry. So show what you would do when you're angry,
Unknown Speaker 17:57
or show this is like a calibration.
Jeff Young:
It’s not before you get to your script, you're like, setting it up. It Up. Yeah.
Sarah Cochran:
So the first step if you want to create a custom avatar is that you have to that you have to record yourself reading a script. Okay, so it's like 90 seconds or two minutes, somewhere in that realm, and you just stand there in front of the camera and you read your script. And the script, at least for Synthesia, takes you through various emotions, and it tells you to make hand gestures when you normally would make hand gestures and all of this stuff. So, I mean, it's, you know, it's interesting, and then you have to try to be authentic when you're reading it. Because I don't know about you, but when I read something, I'm not exactly feeling very authentic.
Jeff Young:
No, it's hard. You're right, if you're reading it, it's, it's challenging, yeah, yeah.
Sarah Cochran:
And, and so that was, that was recorded at my house. I did choose a wall that didn't have anything on it, so it was easy for them to take out the background and stuff like that. But, you know, on my computer's mic, on my computer's video, you know, if you want a much higher quality, that would be like, if you're going to use that avatar quite a bit, going into the studio, being in front of an actual green screen, with a decent camera and a decent microphone, would produce an even better quality avatar. You could fool even more of your friends. Yes, that's what I'm saying, 100%
Jeff Young:
after the break, now that professors can make AI video versions of themselves for teaching materials. Should they Well, look at some thorny issues this all raises. Stay with us. This is where the ad break is going to go for now, I just want to remind everyone to check out the learning curve newsletter. The new installment is going to have a link to this video avatar of Sarah that we're talking about. And you can see for yourself whether you think it looks a little creepy or works. Sign up for the newsletter at learning curve.fm, and click on newsletter. Now, back to the episode. Okay, so I did. I My first reaction, and I'm sure some other listeners might share it, is, I. I don't know, like that, there's something that maybe this shouldn't even be done because you've Yes, you can. And here we are, like, you know, it has these benefits. You can do this now. It was not possible a couple years ago. But, you know, should you automate this? Is this something that could go too far, even if you are not at your institution to going to do that, you know? I know some faculty, for instance, are concerned broadly about AI being used to replace them, like after they leave, say, or if you're an adjunct, you know, and you do this and you make the course, well, why do they need you anymore? So what would you say to some people that would worry about some of these labor issues as these tools roll out?
Sarah Cochran:
Well, I think, I mean, it's a, it's a, you know, it's a valid concern. And having your likeness, you know, it's not your likeness, and then your IP, right? So there's, yeah, your
Jeff Young:
intellectual property is, yeah, it's, that's, and that's why Matt Damon is not going to is not going to do it right, like the actors have have negotiated for very, you know, for control over how AI, or at least discussion of how AI is used,
Sarah Cochran:
Yeah, definitely. And the thing is, with these with avatars, you can create interactive ones, you know. So that's what we're that's what we're building for our website right now is to have a interactive avatar
Sarah Cochran:
like that. I could chat with you in text and you would respond
Sarah Cochran:
your avatar. Yes. Well, we're having a debate on whether or not we want to use one that's like a public avatar, or if we want to use like, whether that be myself or a member of our team, or something like that, as the avatar. So we're still, actually, I wouldn't say, fighting, but having a discussion about who, who it's going to be, because, because of these questions, which is, let's say someone leaves, and then we have your likeness on our website as our avatar that's answering people's questions and and everything else. So in that case, you know, we're leaning maybe more towards a public avatar, because it's it does seem strange if you know what, let's say, I used our student worker and that he was our avatar, and then he left UVA or graduated, or, you know, didn't continue on with us. It'd be strange to have our student worker still being our our avatar answering our question, right? Yeah,
Jeff Young:
yeah. And what is a public avatar? How would you do that? What is that? Is that your mascot? Like, what do you what does a public avatar mean?
Sarah Cochran:
Well, I mean, you, you could create one that's your, your mascot, or I could turn my dog, Rowan, into our avatar, if I really wanted to, of
Jeff Young:
course, I'm not thinking, right? I'm like, of course, it's your dog. That would be it. I My, my imagination has not gone through this yet.
Sarah Cochran:
Okay, oh, yeah, no, no. You mean you could choose whatever. But they also so the tool that we're we're using right now to create the interactive avatar is, Hey, Jen. And so they have, like, just their public avatars that you can use for the interactive avatars, and have them up on your
Jeff Young:
your websites. So it's stock art, so to speak. Or basically, yeah, okay, so if this keeps going, in five years, is it, or in some small amount of time, do you imagine there won't be professors making videos anymore, because this will be the way to go. This is avatar those
Sarah Cochran:
professors, are going to have to make the videos still, because I think you're going to have the reaction from similar to my student worker, which is just wanting to hear their professors, their stories and their context, and how this knowledge that they're imparting they've used in their own experiences, like the relevancy and and whatnot comes from, from the professor. And there is something about in talking to a person, we're talking virtually right now, but like, I can read your emotions. They seem very, you know, very realistic, except for that one time when you froze.
Jeff Young:
And I, like, I might have to mention that into things real anymore. It's probably because you're, you're so steeped in this that you're looking for like, oh, wait, did I lose my human
Sarah Cochran:
Yes, right? Is my human God? Now this is terrible, but you, you, you feel that connection even virtually, right? And so that's still terribly, terribly important. And I don't think that we're going to lose that.
Jeff Young:
So UVA is in the very early stages here with lots of questions still to sort out. And after talking to Sarah, I had more questions that I was curious to ask. So I had a longer call with Derek Bruff to hear what he thought of all this, because these days, he is a national expert on AI and teaching, and he's currently co writing a book called The Norton guide to AI aware teaching.
Derek Bruff:
I think one thing to keep in mind is that, certainly right now and probably in the future, we need to be really transparent with our use of AI with our students. There was that story earlier this year. Maybe last year, I think there was a student at Northeastern University
Jeff Young:
who was, I don't I'm sorry, I have to stop you. The northeastern student is the last episode I did. I chatted with her.
Unknown Speaker 25:08
Oh yes, you got to check it out.
Unknown Speaker 25:11
I haven't the queue.
Jeff Young:
I feel bad. I interrupted Derek there, but I was just so excited because he was about to cite the situation I covered in our previous episode where a Northeastern University student caught her professor using AI to make slides and lecture notes without telling the students that it was AI, and she lodged a formal complaint with the university. People keep raising that yeah, scenario, because it is. It's very specific. She asked for her money back because she felt cheated.
Derek Bruff:
Yeah, regardless of the specifics of that scenario and how it turned out, I think there's a general principle there that, yes, our students are expecting us to show up as their teachers, and what showing up means can be very different, right? So let's take the flipped classroom example, and let's think about I'll also caveat. I flipped my classroom for years without using any video work whatsoever. I would have my students read the textbook. I would give them very targeted readings with guided questions and some questions that they would answer for me. And so, like, there's a way to structure the flipped classroom that is not video intensive. I just want to put that out there. But many faculty do find the video as a useful medium for that pre class, exposure to whatever the content is, right? Give the let
Jeff Young:
give the sort of like things you need to know to then do something with the material, yeah, that the student may not be familiar
Derek Bruff:
with, and certainly in a fully online course, right? Especially an asynchronous online course, the kind of video delivery of lecture material like there's, there's always going to be some need for that. I think it's interesting that some faculty I know who flip their classroom with video, don't make their own videos. They find other people's videos that are quality, right? They curate high quality, you know, educational materials for students to encounter, right video or otherwise and so, and
Jeff Young:
students might not even object to that. Probably, I don't hear, you know, if it's a TED talk, people use TED talks in in person or online. It's like, this is really a fitting moving, you know, 16 minutes of jail.
Derek Bruff:
And there's plenty of faculty who like to create open educational resources for this purpose, for their colleagues at other institutions to use. So like they've, they've released, you know, whole sets of videos around commonly taught courses. And so I think where it gets weird is when we have a fake version of ourselves, right? It's, it's actually cool to have videos from someone else, right? That like, we don't think that's weird. Some students might object. They might say, Well, why aren't you making these videos yourself? I can imagine some objections there, but if the rest, that's why I said it's important to show up. So if I have my students watch some videos from, you know, some faculty member at another institution, and then we have class together, and I help them make sense of what was in the videos. We do great in class activities, right? I have a strong instructor presence, even if I'm not making all the content myself. It's that instructor presence that I think is really important in an on site class, but especially in an online class, and there's lots of different ways to show up. And so I think, I think what would be problematic is if the only way I was showing up for my students was through my video presence, and then I replaced myself with a an AI avatar, like I've already got a fairly limited set of tools in that case, to have some type of relationship with my students, I don't want to, like, throw out my major tool to do that. And so having this type of resource as a supplement to other ways that you show up for your students, I think it's going to kind of be a necessary condition here. Otherwise, students are going to feel gypped right that they're not getting something
Jeff Young:
AI video has been in the news in the past couple weeks, with the release of Sora by OpenAI and other similar platforms, where users can quickly make AI versions of themselves, and by just typing a few instructions in a prompt, they can make their avatar do all kinds of crazy or funny things that technology isn't about teaching, but it sparked discussions Of all kinds of potential misuses, like deep fakes, spreading misinformation using these tools.
Derek Bruff:
I won't name names, but there are certain politicians in the United States that like to create AI videos of themselves doing things that they themselves would never do or could not physically do, perhaps, and so. And you know, you look at some of these AI videos and you're like, that's clearly a fake, clearly that person did not do that thing, right? We know it's a fake, but it is received in very interesting ways by different audiences, right? Some folks love this fictional version of their favorite politician who does these. Super heroic AI generated things, and that they have almost a relationship with the fictional version of the person, not the actual person, what they actually can do or what they've said. And so I think that's a little problematic, right? I think if we're fully aware of that, right? Like, I'm a big fan of Star Wars characters, but I know they're fake, right? I know they're fiction, right? I don't get confused there. And so I think, I mean, I can imagine some very clever faculty who are again, connecting with their students in a lot of different ways anyway, and they have a little fun with their their AI version of themselves who does things in the course, right? And maybe they all cut, you know, we can sit around and kind of mock the AI version of me with my student, yeah?
Jeff Young:
It gives a punching bag. It gives us like a fun little Yeah, straw man there. But
Derek Bruff:
I also know faculty who who would not want to do that, but would be comfortable creating AI videos that are not them, right? So, so I think one of the use cases that I see a lot, a growing use case for AI in general, is to put students in some type of simulation. So this type of pedagogy happens a lot in nursing. It happens in education. Happens in some other fields where you want students to have a chance to practice talking to a patient or practice talking to a parent of a student in your class right?
Jeff Young: 31:23
Right before the stakes are real. Just have this, yeah, safe environment,
Derek Bruff:
right? And like, if I think about kind of medical education or nursing education, sometimes the simulation is a robot dummy and you're practicing physical skills, right? That's a different kind of use case. But other times, they have trained human actors who will play the role of the patient or the family member or what have you. Sometimes the instructor will have to play do that role play in order to make the simulation kind of authentic. Sometimes it for students to role play with each other, right, where one gets to be the patient, one gets to be the nurse. But you know, all of those are kind of approximations of the real thing, and they have kind of pros and cons. So here's a way to do a simulation with an AI that is going to respond in some ways that are directed by whatever the prompt is happening or whatever prompt it's given. And you know, the easy way to do this is text only, right? There's lots of tools right now that you can do that. To create a interactive video simulation takes a little bit more technical know how? But it's not impossible. There are institutions that are looking at doing in this and so, so then you think, Okay, in this case, the video element might add a lot if it's done well to that simulation. It's probably more like the real thing than a student doing a simulation with a peer. It may be less like the real thing than a trained human actor who does this regularly, right? Like there's anybody who has that actor to hire, but that's a resource, right? That Some institutions have that type of resource and some don't. And so if you think of this as kind of a constellation of options for trying to do a simulation, then you can start to see, okay, well, this could have some utility in some cases and not in other cases. And in that case, it wouldn't have this added effect of somehow replicating the instructor, like you wouldn't have that, that weirdness to deal with, because everyone knows it's a fictional character that you're interacting with, the quality the interactions are still interesting and useful educationally, but you're not doing anything to disrupt the relationship the student would have with the instructor. I
Jeff Young:
like your your kind of, you know, positioning it as showing up, and so as long as there's some acceptable to the students and to the to the employer, degree of showing up, whatever. Any of this could be fine. And I, what I also think is interesting in that example you've switched to with this, you know, a video simulation, whether it's in the character of the professor or not, it seems like it that clearly feels like something new and an addition that's not, you know, just being the lecture. I think that's the interesting thing that, you know, we're bound to see as AI matures, instead of just putting it in to do the thing that has always been done, adding something or trying some other new way to involve this technology that wasn't possible before because of resources or whatever.
Derek Bruff:
Yes, I think this is, this is a hard thing to grapple with, though, because sometimes when we have it do something new and different, we react strongly and negatively because it's new and different and weird, and because I hear a lot, there's been a lot of work over the last couple of the last couple of years in education, of trying to think of, like, what are the metaphors we use to think about AI? And there's a lot of different metaphors, and the different metaphors bring different lenses, right?
Jeff Young:
And so I'm a fan of this, yeah. This, this thought journey, yeah. So, yeah. So, okay, so these metaphors, how does it fit
Derek Bruff:
here? Well. And I think about like, let's say. Another use case I hear a lot about, we want to give our students some type of AI powered tutor bot, so that when they have a content question at two in the morning and the instructor is rightfully asleep, they have someone to ask, right? They don't have to wait till the morning. And, you know, sometimes instructors even kind of style these tutor Bots as like a robot clone of themselves, right? Sure, the AI twin, right? The AI twin. And, you know, maybe it's just text, maybe it doesn't have the video piece, whatever. But I think it's interesting to think, what are the students? How are the students using this AI tutor bot, and what would they have done if they hadn't had that available at 2am so you could compare a student interacting with a tutor bot as like, Well, gosh, they could just come to my office hours and ask me the questions. I would love to answer their questions and have that connection with my students during office hours, and the tutor bot is clearly an inferior copy of me, so like this. Like, why would I want them to do that at all? There's no reason. But if you compare the tutor bot to the student googling an answer at 2am on their phone, then you start to see, well, maybe that's a better option for students, right? I can train the tutor bot to limit itself, maybe to the content of my course, to pitch the content at a level that students can understand, and maybe even train it to kind of be an authentic tutor, where it has a kind of back and forth and an athlete in questions and prompting questions and does more instead of just giving away the answer, instead of just giving away the answer, right? What Anna Mills calls an unethical tutor, where you just answer all the students questions, right, right? And so, you know, in that case, so a lot of it is what you compare it to. If I compare it to a replacement of an interaction with me, then it's problematic in some ways. If I compare it to a replacement with an interaction with Google, it may actually be a much better option. And so that's why I feel like sometimes I love thinking about, what can we do with these tools that we've never done before? What kind of new learning experiences do they come up with? I find with AI, people have such a visual visceral reaction to the technology that I need to situate it more often among things we already know and help people kind of situate it usefully in that landscape.
Jeff Young:
So coming back to my initial reaction to these avatar Professor videos, my first reaction to the idea was, again that robots were taking away a key human aspect of teaching. But an interesting counter argument is that lecturing when it's just broadcasting information is not really where the important human aspect of teaching rests, and if that, our choice is between a pre recorded video lecture with a professor versus lots more up to the minute videos where a bot version of the prof is made by a human professor. Maybe that is more human, especially if that means the human professor is using that time to be talking and working more directly with the students. I do wonder, though, whether that is how this is all going to play out.
I am curious to hear what you think.
Please send me reactions or your own experiences to Jeff at learning curve.fm as you see, I do read these messages, and it might even spark a future episode.
Jeff Young:
This has been learning curve episode six. Please tell a friend or colleague about the show to help spread the word or share it on social media.
You can get show notes and find out how to be a paid supporter at learning curve.fm and you can sign up for our newsletter there to keep up with the latest this episode was written and produced by me Jeff Young.
Our music is by Komiku, and the episode art is by mid journey. We plan to be back next week with a special episode made in collaboration with student journalists at the University of Minnesota. Stay tuned for that. In the meantime, thank you for listening.
You you.