S4 E29: Sensible Approaches to Teaching in the Age of AI-Generated Everything

In this episode, Dr. Tawnya Azar joins your host, Dr. Rachel Yoho, to share actionable strategies about how to avoid becoming the “AI Police” and teach students to think critically about how and when to responsibly use AI in college. Additionally, she shares how to meaningfully build rapport and trust with students in an online classroom and provides insightful tips to help students find value in demonstrating their learning – rather than relying on AI. Listen in to find out how Dr. Azar found joy in teaching in the age of artificial intelligence! 

Resource: Yoho, R. (2023), No, Let's Not Go Back to Handwritten Activities: Inclusive Teaching Strategies in the Context of ChatGPT. The National Teaching & Learning Forum, 32: 1-4. https://doi.org/10.1002/ntlf.30379

Check out our website!

Transcript

Rachel: 0:08

Hello and welcome to the Keystone Concepts in Teaching podcast. I'm your host Rachel Yoho, and I'm very excited for our conversation today. We're gonna be talking about how we teach, particularly asynchronously, within the context of all of these different artificial intelligence or AI emerging technologies. So I'd like to invite our guest to introduce yourself.

Tawnya: 0:34

Hi, I am Tanya Azar. I am a term Associate Professor of English in the English Department, and I teach primarily Composition and an Honors class in the Honors College. My research primarily focuses on online writing instruction pedagogy, community engaged pedagogy, and a celebration of student writing events.

Rachel: 0:57

Yeah. That's great. So thank you so much for joining us and thank you so much for your time and your suggestion. I'm particularly excited about this episode because we are very much due for another episode on AI, so to get us started, can you tell us a little bit about your classes and what you teach that lead into our conversation?

Tawnya: 1:18

Sure. So I primarily teach English 3 0 2, which is Advanced Research and Writing. And as I said, I teach an Honors class on community engaged pedagogy as well. But primarily I teach the Advanced Research and Writing, and it's all asynchronous in that context.

Rachel: 1:35

So over the last few years, I mean certainly AI technologies have been around for a while, but particularly since 2022 with the launch of Open AI's Chat GPT, and the massive explosion we've seen across different areas. You know, what are some of the challenges that you are seeing right now with AI, particularly as we're digging in here about teaching asynchronously?

Tawnya: 2:01

Yes. In the past two semesters, I noted a few major sort of areas of challenge for me as an instructor of asynchronous writing and research, and I set out to address those challenges over the summer. But essentially, it boiled down to three major things. The first one is that I felt like I was spending most of my time being an AI police. Just constantly, constantly noting where I saw it, how I saw it, what it was doing, what it wasn't doing. The second challenge was the hallucinated sources that students were using or hallucinated information from existing sources, hallucinations in general synthetic disinformation. And the third challenge was making sure students could meet the learning outcomes to demonstrate their learning in an asynchronous context, given this technology in play. So those three challenges led me down this path.

Rachel: 2:56

Yeah, I mean, I think that these are such great points and often, especially in some of the workshops and conversations I have with faculty, it's really about one of those key things is that first one you mentioned, you know, it's not always the best use of our time to be enforcing or trying to determine or scrutinize or side by side comparison of student writing or I can tell this is AI. If the programs, if the AI checkers are no good, are we? I mean sometimes it's completely obvious, perhaps. And we can, you can tell the cut and paste kind of things. But I think here really looking at, especially as you get into your third point there, really looking at how can we be more thoughtful? And I wanna just have a brief aside about the hallucinations. So when we're talking about those, for listeners who might be a little less familiar with AI hallucinations, we're really talking there about information that's basically words in an order. And so can you tell us a little bit more about, you know, how it's similar, what you're seeing this as in your classes, as that example with the hallucinated information being"sounds real, isn't" kind of thing?

Tawnya: 4:01

Yes, definitely. So one of the things I noticed was the difficulty I was having, determining whether the student used an AI generator to create a citation that was bad, or they used an AI to create a source that didn't exist, that was cobbled together from other source types. So sometimes you get a citation with an author that existed and a title that existed and a DOI link that existed. But the three things didn't have anything to do with each other, other than in this citation. And so sometimes they would say, oh, I'm just using a generator and it was wrong, which is probably not the case. So my biggest challenge was tackling that issue. And then also there is the matter of synthetic disinformation, where even if you're using a existing source, it can hallucinate facts from that source. Definitely it could, at the very least, it sometimes skim reads it and misses important things, but sometimes it manufactures things, quotes, facts, stories, et cetera. And so trying to help myself and the students navigate that successfully was a major challenge.

Rachel: 5:04

Yeah. I mean, and that also kind of goes along with some of the other issues coming out of different AI technologies. I mean, you can get things along the lines of hate speech created, or you can see some of these others that basically come out of those root causes, if you will.

Tawnya: 5:19

Yes.

Rachel: 5:20

Okay, so with that, that's a lot of challenges. So where do you wanna dig in next? Do you wanna talk about some of the things that you tried? Do you wanna talk about maybe framing this in terms of opportunities or where do we go from, wow, these are big problems?

Tawnya: 5:35

Yes. So one of the things that a lot of my colleagues are trying to figure out is I am not an AI user, so how could I possibly teach students how to use AI in a responsible manner? And I think that actually I would leave that for the end because for me, I started with what I call the low hanging fruit issue, which was how do I transform these very specific modules, these very specific activities in a way that releases me from the obligation I was feeling to be an AI police? If I could get around that, then I could maybe conceptualize an approach to teaching them AI in a way that made sense to me. So I'll start there and that first hurdle was how do I maintain the integrity of peer review, in particular, at the time the students were taking each other's work and throwing it into AI. Taking the feedback AI gave them and throwing it into the LMS and I, it just defeats so much of the purpose of having students peer review each other's work. So,

Rachel: 6:35

Yup! Yeah, it does.

Tawnya: 6:37

Um, so I decided that I would use a framework called TAPs, which is Think-Aloud Protocols and adapt it for the asynchronous context. And what I did in this case is I have the students read small snippets of each other's work and record it. And then they record their feedback on that piece of their peer's work. And even though of course they could still be using AI to help them with that feedback component, they still have to actually read each other's work, which is half the value of peer review is having them read it, and absorb what their peer is writing. And also the upside, uh, the sort of pedagogical bonus, to this approach too was having the peer hear their work being read by someone. In a similar way, we have writing center tutors that read work out loud. I think this was sort of a bonus, that I wasn't even thinking about when I first designed it, but definitely something that emerged as I was listening to their audio feedback. So that was sort of the first hurdle that I tackled. And the second practical low hanging fruit hurdle was how do we get around this synthetic disinformation issue? And this, I'm all going toward the point of not being an AI police anymore. And one of the ways that was happening was trying to figure out, as I said, is this a bad citation or is this an imaginary source? So what I have them do now is I have them submit PDFs of all their articles in a OneDrive folder. And then at the very least, I can track down the source they're setting in their writing and see is it in their folder? Does it say what they say it says? And that was just a huge time saver for one. For another, I think that act of having to put the PDF in the folder helped just avoid some of the major temptations that students have to kind of get around that entire aspect of researched writing, because they're all working on very different topics in these 3 0 2 sections. It's not like we're working from a common set of texts that I would know ahead of time. Again, one of the pedagogical bonuses from doing it this way was it did actually allow me to have more meaningful conversations with them in our research meetings because I had access to all of their sources, you know, that I could look at ahead of time or in the meeting and then dig down to see what they were able to recall from those, how it compared to what was in their research spreadsheets and just really kind of go into it with them in a way that I see now as really helpful to them in writing the literature review that they have to write. Because I was asking the questions I wanted them to address in that review. And I think that conversation we had was better because I had access to the PDFs at their sources. So those were sort of the two initial hurdles that I was trying to clear is those two major things. And then the third one, the AI police is all about kind of making space for the conversations I wanted to be having with students about their writing and their research. Taking AI essentially off the table, so that we could have conversations about writing and research. And so far it's going really well. I just can put that out there. I mean, I haven't studied the class in detail like I'm going to. But instead of, you know,"You're over relying on AI. You used AI here, I can tell" this sort of thing. Instead, it's more like, Oh, because it's not banned entirely in this class. I'm going to use it for the things she said I could use it for and not overuse it, because she's very aware of what that looks like."'Cause I show them, and we can now talk about these other things. I got an email from a student the other day about how, even though all her sentences are grammatically correct, which bonus of AI for sure is that most people's, most people's writing are grammatically correct.

Rachel: 10:25

That's good!

Tawnya: 10:26

Now, um, she's like, the feedback from my peer is still that they're run on sentences. And I'm like, oh, okay, well maybe the issue isn't the grammar. It's that it's the structure of your ideas. Let's talk about that. So it was just really nice to have those kinds of writing conversations instead of,"oh, I didn't use AI here. I, you know, I didn't use it this way. I didn't use it that way," and so on and so forth.

Rachel: 10:48

Yeah, I think this is great. So I wanna kind go back to what you were talking about in terms of where the students can use AI and not over-relying. And so a lot of the things, I mean, I have an article we've talked about, I think on the podcast before about, you know, not going back to like all handwriting, you know, it's not productive to just say, okay, we're gonna ban these things, but how do we look at this as a tool? And so can you talk to us a little bit more about how you communicate with the students about when and how to use a tool. We don't wanna assume people know things, right? So what does that look like for you in practice?

Tawnya: 11:28

This is a great question because one of my passion areas is digital literacy with students. I think they get thrown the book a lot and you know, with very little instruction. We assume because they grew up with digital technology, that they know how to navigate it in professional context was just not true.

Rachel: 11:44

No, no, no,

Tawnya: 11:45

Uh, no. So one of the things I emphasize to them is that learning AI is supposed to be part of a suite of digital literacies that they must consider, if not practice, while they're pursuing higher education. So that they are more comfortable using it or not in the workplace. And one of the ways I do this is I do start with an orientation to show, to go over with them what exactly AI is and what it isn't. And then also to articulate, you know, the various ways I'm gonna show them how they might use it, if they want to, and the ways in which they're prohibited from using it for the purposes of the course. And then throughout each module I will always have, even if the lesson has very little to do with AI, I'll always have a prompt that says, can I use AI for this? And then I answer the prompt and I say, you can use AI for this task or this task, but not for this task. A lot of the not for this task are related to their perspective on something. And so I emphasize that I want their perspective for that particular part of the activity. And I also emphasize throughout the orientation and throughout certain parts of the course, you know that AI can create writing faster and more grammatically sound than most of us can, but it doesn't necessarily produce better writing than my students are capable of doing. One of the things I hope they come away with is sort of a boost in confidence from those lessons. And then the only other way, I sort of integrate AI into the course is I do show them a few times, like, here's this common agent that, for example, our document agent that Mason Patriot AI has. And I show them how it took what I gave it, and answer the prompt that I prompt my students with and show them how it. Was okay doing this part of it and this part of it, but not so great at doing this part of it so they could kind of be aware of its limitations and when they go to use it, then I'm more likely to get a quality output from them because they know where the hangups are.

Rachel: 13:47

Yeah, absolutely. And there are many different tools and I think that's a really important part of the conversation. I know you highlighted there, you know, one particular thing through Patriot AI and we certainly have. Lots of opportunities within even just the mason computing kind of ecosystem there. But I mean, I, for example, I was doing a talk recently, and as we're talking about this in terms of tools, it's not only knowing what to use, but when to use it, right? So when we do this, you know, I could say, okay, well there's a tool room over there. Go get me something. Well, something doesn't help, right? I mean, I could say, okay, go get me, you know, the particular tool. But it might help if I describe what that tool looks like. Right? So, and I think that kind of goes along with what you're talking about there as well of not just, you know, when and how to use in the course, but where are the opportunities? Where are the times and where are the times that, I think it goes back to that overarching concern you were talking about earlier of, you know, the academic integrity of the course. You know, a lot of that is that critical thinking piece. So not outsourcing that right. So you've been doing some really cool things, you know, these are really fascinating almost like interventions really for asynchronous instruction specifically. I think all of these that we're talking about also crossover really well, even if you had these activities as part of a synchronous course. Right? But that is an even more challenging environment to teach in. But with that, can you tell us a little bit more about your discussion about AI and the value, I think hear more about the value of writing instruction.'cause I think that not only maybe crosses over with your teaching, your research, you know, where all of this comes together for you.

Tawnya: 15:29

Yes. So in the first week of the class, the student's first discussion is about the value of writing instruction in the age of generative AI, and I do have a few optional readings linked that they could read if they want some additional context. One of the reasons I set this discussion is for one, I kind of like to unpack with them some of the rhetoric around AI and especially in a writing context and in a higher ed context. And this is one of the ways that I think I build rapport with students, as I say, and I think very vulnerably, I say, I don't know what to think about this yet. Like this is still very new and I am not certain, and I don't think most critical thinking people are certain exactly what this technology is going to do, how long it'll be around in its current form, what impacts it will have in not only higher ed, but in the workplace, in our personal lives, and so on and so forth. And I want them to think about and to be honest with me, if they have particular thoughts on this. You know, what value could they possibly see getting out of a course like mine, especially given a tool that could, in theory, do all the things that I asked them to do almost effortlessly? And so I think it was, it's a bold move to, to challenge the value of your own course in the first week at least.

Rachel: 16:51

Sets that stage. This is gonna be that course that really pushes that. So, no, that's great.

Tawnya: 16:56

And they were, they were honest. Certainly some of yeah, AI can do this all. And some of them were like, no, AI can't do this all. And some of them were like, I don't know either. You know? But I think it helped them and me to have a relationship from the start that is like, I am in this uncertain space. And I tell them very clearly in the orientation that I do not wanna foist any moral approach on them to this. Writing for me comes incredibly easy. I would not have gotten a PhD in English Word otherwise. Um, I'm a high resourced, very adept writer, but that is not the case for the vast majority of people in the world. And so, I tell them, I don't begrudge you the use of any tool that will make this task a very difficult task, easier, but you need to demonstrate that you're learning what I'm teaching you or I can't give you credit for this class. So that is the main thing. And I think kind of taking that morality off the table, that judgment off the table really helps them see it in a way that's like, oh, okay, yes, she needs to see that I'm learning. That's all, you know. And while I do encourage them to think critically too about the environmental impacts of ai, the privacy issues and so on, I link to readings on that. Since AI is not the focus of my course, I don't, you know, lean too heavily into it, but I do want them to think about it as part of their choice making about whether to use it or not. And I do not require their use of it at all. But I do think framing it in that way of you need to show me that you're learning eases the pressure and the uncertainty they face in a lot of other classes where they're just scared to death of failing the course or getting kicked out of the institution because they willingly or not used an AI technology, and whether and how to do that. So, so far anyway, that is my take on how I'm approaching it.

Rachel: 18:46

Yeah, I mean, I think that that sounds great, especially when we have the broader context, right? We have in the national conversation there's constant questioning right now of the value of higher education, the role of higher education. You know, we've seen in the last number of years that in a lot of spaces in the higher ed spaces, there's discussion of teaching becoming more transactional. And so that, I think helps even pull out of that space, what are we doing? Why are we doing it? How do we make this valuable? How do you demonstrate your learning so that you get the things you need? I mean, the academic integrity, the student engagement, that's all deeply intertwined there, you know, in that conversation and having our students engage maybe not with some of those topics that I just mentioned, but especially the interactions there I think is a really powerful way to, to kick off that conversation. I'm, I'm fascinated. Yeah.

Tawnya: 19:40

I'm still learning from them about the best way to go about this and very open to tweaking my own approach. But I will say this is the first semester in a number of semesters that it felt less like a burden and more like a joy.

Rachel: 19:55

Well, that's great. And that's, you know, that's part of teaching, right? We try it, we see it, we try again, all of that. So yeah. I think this sounds great. I mean, we've had some really great topics that you've talked about. Is there anything else you'd like to share with us?

Tawnya: 20:11

I think, setting AI aside for a moment, the one thing that I want to share about asynchronous pedagogy is that I do think the payoff of building rapport with your students, which is very difficult in an asynchronous context, is just immense, you know, building that rapport through an appropriate amount of vulnerability with them, offering as much flexibility and choice as you can in a pre-designed course, you know, showing them that you are actually a whole person and you are looking at their work and you are addressing their concerns, whether it's privately or via class-wide announcement. You know, those types of things go a long way to, I think creating a more valuable learning experience for students. And, you know, I just wanna challenge the set it and forget it sort of approach I think a lot of people might take to asynchronous teaching. This is a passion of mine and I think students deserve to have an enriching learning experience regardless of the modality.

Rachel: 21:08

Absolutely. I agree. I've taught quite a few fully asynchronous courses and it's amazing the types of connections, the types of instructional relationships that you can build with the students, leading into very fulfilling professional trajectories and all of that without having to sit in the same four walls. You know, when we think about the limitations, I think like you said, the concept of it, the set it, forget it or, you know, how do we interact, and also from the other side that, you know, I am not an AI bot grading your stuff. Like I'm an actual person here and I'm giving feedback on this. And how do we build that through the semester? Absolutely. Yeah. So I think these are such great points. And it's so important. I mean, whether we're teaching asynchronously or synchronously or in-person or some other version, looking at and applying evidence-based practices is important, but perhaps the most important in fully asynchronous classes. What's really coming to mind for me, for our keystone concept is really looking at academic integrity. Part of our conversation today was that we're not just looking at this from the perspective of catching cheaters. That's not just reframing the conversation, but throwing that out and starting over of how do we have our students learn? How do we show them that we want them to demonstrate their learning like you were talking about? And really have this be the integrity of the course, our academic and professional integrity and our students' academic and professional integrity as well. Like you said, they're not showing up in our classes knowing exactly when and how to use whichever tool, and so how do we create those environments where they do learn when and how to use those different tools.

Tawnya: 22:58

Absolutely.

Rachel: 22:59

So, yeah, yeah. So with that, you know, thank you so much for your time. I really enjoyed this conversation. I think we have not only the important space of AI, but the important space of asynchronous instruction, but things that also apply across our different modalities. So thank you so much for joining us.

Tawnya: 23:16

Thank you for having me.

Rachel: 23:18

Yeah, and please catch our next episodes. We're posting every two weeks on Keystone Concepts in Teaching during the fall and spring semesters, and you can find us on most of your favorite podcast listening platforms. So thank you and I look forward to catching you in another episode.