Episode Transcript
[00:00:00] Speaker A: Welcome to FHSU Tilt Talk, a podcast about educational technologies, teaching and learning, scholarly research and service hosted by Teaching, Innovation and Learning Technology. Staff.
[00:00:15] Speaker B: Welcome to our special series where we dive into Generative AI at Fort Hays State University.
In this series, we interview faculty and staff about their work with Generative AI, exploring how they've integrated it into their teaching and research.
Thank you for joining us for our Telk Talk episode today. I'm Magdalene Moy, an instructional technologist in teaching innovations and learning technologies and the chair of our Generative AI Task Force at Fort Hays State University. And I'm joined today by.
[00:00:49] Speaker C: Hello, I'm Dr. Gary Anderson with the Advanced Education Programs Department.
I'm also the coordinator of the Transition to Teaching program and a member of the Generative AI Task Force.
[00:01:04] Speaker B: Today we're interviewing Dr. Matthew Clay, an assistant professor in teacher education, to talk to us about his use of Generative AI in his courses. Matthew, would you be willing to give us a brief description of how you've been using it in your courses?
[00:01:23] Speaker A: Sure. So most of the courses I teach are methods courses, so some of the main focus is on elementary science teaching methods. And then I teach a course that is for all secondary education majors.
And so the main use I focused on is in creating examples for class discussions. So in the secondary school course, one of the long term goals, you know, big picture goals, has always been how do we prepare these students who are very close to having classrooms of their own to deal with a lot of the, you know, the things that just pop up that they're not on a lesson plan. We don't have a class for it, but it pops up and it can make a teacher's life a lot more difficult. And so one of those is dealing with upset parents. And in the past I've always had a few examples of kind of generic emails that an upset parent might send. And I would sit down and write those and you know, they would be things like, oh, I'm upset about their grade or I'm upset that they got a detention.
And what it's been really helpful, and I've used ChatGPT for this, is that now I have tailored those examples towards the content areas that the different students in my class are teaching it. And so we now have examples specific to music class and examples specific to PE teachers which look a little bit different than what I tended to think of as a former middle school and high school science teacher.
Also in those, we've been able to get some examples of things that, to be honest, were really a Bit difficult for me to sort of. Right. In the kind of, like, nuanced manner. So one of those is a parent that's accusing a teacher of racial bias, which is, you know, that's. That's certainly a challenging situation. That's something that I would rather we have that discussion in my classroom instead of the first time that possibility comes across. Their mind being in their own classroom by themselves, opening the email at the end of the day and going, oh, my goodness, what do I. What I do about this?
And so it's been really helpful in generating those examples.
The kind of beautiful thing of it is you can get the specificity of a particular content area or a particular challenge incident that I want described, and you can generate those.
I mean, I. This year I probably went into class with 20 examples. Some of them we didn't use. You know, some that as you're going through class, you're like, well, we kind of discussed this one on the side anyway, so we'll skip it. But it generated those 20 examples in the time that used to take me to write one.
And the one was never as specific.
So that's one use. Another use is that within my elementary science methods course, a challenge is that although they write lesson plans and create lessons, a lot of times when they go into classrooms, they are given materials to use. And so we're trying to have a discussion of, like, how do you look at a lesson plan or a curriculum critically and think about, what am I going to change? What shortcomings are there?
And a good challenge to have, but a challenge, especially among our elementary ed majors, is that they are very nice people that have a difficult thing saying something critical about someone else.
And so the first time I used ChatGPT to generate, I think, about 12 sample lesson plans, the prompt I gave was just generate a lesson plan that aligns to. And then gave a standard number.
And the first time I used those in class, I passed them out. I didn't give them any context. I just said, I want to know, would you teach this? Would you not? What would you change?
And so they, they had little details that, like, oh, I think this could be a little bit more clear. I. I think this part is maybe a little bit too long. You know, these, like, little tiny changes. And then I kind of broke the, like, okay, these weren't. And, And I told them, I said, I didn't write these. I didn't write these. Per the. The writer is not here. You're. You're not going to offend anyone. They had these Little changes.
And then you said, okay, well these, these are all AI generated. And then they started tearing into them where they're like, actually this assessment doesn't match this objective. That needs to be better. These. And it was when you're like, wonderful, like, that's how critical you should be of any resource. Because even if a human wrote it, they didn't write it for the kids in your classroom.
And so having the non human created example that, you know, these students who are just really too nice to be hurtful of a human, they had no problem like, really being critical in a way they need to be of these AI generated. And it also was a nice example in the sense of you're going into this career field, you have to bring something that AI can't do because it wrote 12 lesson plans while I was eating my lunch, you know, and you, you need much more time than that and, and want to be paid for creating lessons.
And so it ended up being a great lead into a discussion about like, intentionality and okay, what, what do you know about your students and how can you be intentional about matching their needs, interests, etc.
In a way that I don't know if that discussion would have happened without this sort of like, counterexample of, hey, here's what the machine can do.
So you have to do something more than that.
And those have really just been, I think now, I guess, three semesters I've used those. We did all started last spring.
And so that's each of those activities, you know, it's been iterations, but now three semesters we've tried them.
[00:08:42] Speaker B: That's fantastic. I like that challenge of like, what can you do better than the machine? And that being a way for them to be like, oh, I can be mean now. I interested, though, to know with your first example where the AI generated emails were coming from parents, they knew they were AI generated.
[00:09:04] Speaker A: Yes, they did. And, and that actually became. Because we had a little bit of a side discussion by. If the email really looks like this, that parent's probably a lawyer, and you may want to be careful about how you reply to it.
But that actually ended up being a nice discussion of. I made the point of. So that you, you all don't care that these are AI generated. Like you care that I thought about.
I want examples relevant to your specific content areas.
But they're like, no, we don't, like, we don't care that it's AI generated instead of you setting down and spending four hours writing sample emails. So, yeah, those. It was very transparent about. This is where they came from, this is how I did them.
As an example of, you know, for your future students, the intentionality and how you approach using the tool is what matters more to them than the tool you used.
[00:10:06] Speaker C: What has been some of your students reactions to these assignments? What have you noticed that the students are saying about it?
[00:10:19] Speaker A: One thing that has surprised me kind of in general about their reaction to AI and it shifted a bit over the three semesters, but they seem really scared or skeptical of it as a tool in the sense that they're like, oh, this is cheating.
Even in cases where you're like, okay, but it's, it's quite obviously not like you, you're, you're using spelling, you're using spell check. Like you didn't think of that as cheating.
The, the discussions that have come out of it have been really positive. And the, I think the reaction to me, what I've noticed is more of oh, wow, it can, it can do that. And then there's even been some like, okay, well can it modify that lesson plan for blank?
Which has been fun because you get to be like you say, let's try and you know, type in the prompt of modify the above lesson plan for.
And so it's, I think it's been nice for them in being able to see the process of just kind of exploring and using a new technology and not a, oh, I know all and have mastered the use of this and know exactly how it should be applied and what I'm doing and instead being able to see someone model, oh, I don't know but I think there's potential here.
[00:12:02] Speaker B: I'm just curious whether or not it's probably not super important, but are your students face to face students? Are they online students? Are they graduate students?
What are, what's your student population for both of those courses that you mentioned.
[00:12:18] Speaker A: Those were both face to face activities and undergraduate.
The elementary science methods I do teach online, but we've not adapted that activity since it's a class discussion. I've not adapted that into a discussion board type activity. But as I sit here, that's when I'm like, oh, I maybe ought to think about doing that.
[00:12:49] Speaker B: Yeah, I'm just curious, like you know, having someone in, you know, who can look over your shoulder and kind of see, you know, what you're using. Although I guess they're not really using the generative AI to create the prompt. You're using it more as like a content around the discussion, like how to prompt it with Your students rather than having them actually use the tool. Is that correct?
[00:13:12] Speaker A: Yes. Yeah. In these examples, yeah, that's the case.
[00:13:17] Speaker B: If you put it in an online setting, please let us know how that goes.
[00:13:23] Speaker A: Yeah, I do. And I, I do think that's when like, you know, more meaningful discussion board type things is always one of the like for me, the areas that I feel like can be improved and never quite content with.
And so that now I'm like, oh, I, I need to think about that. It gave me something to do for the summer.
[00:13:44] Speaker C: I'm just curious about where do you see this going?
Like Magdalene said, this is, you know, you're generating a static document and then they're responding to it in a discussion. And do you see some more like iterative based actions in the future with this kind of AI generated content, maybe interactive?
Where do you see your use of this moving forward in the, in the future?
[00:14:22] Speaker A: I think one step is with pre service teachers is being more intentional about outlining sort of best practices for them, you know, as we figure out what those best practices are.
Because it is one that like I said, they, they struggle with.
And I've even had some where they said, no, that's, that's cheating. I'm saying like, no, like I'm the one you're giving the assignment to and I'm telling you, like, use this. So like one example was they had a student who, who was struggling with her writing a bit and said copy and paste your paper in and ask for feedback. Right. Like you, you can do that at home. That's, that's all tools available to you. It's like, no, that feels like cheating. Am I, it's one, I, I, I'm the one you're giving it to. Like it's not, it's not cheating.
But I think being more intentional and teaching those practices is where it's really going to head. And I, to be completely honest, I think there are some skills within our program we're going to have to look at.
Is this still a relevant skill to teach or does this skill now become editing AI generated content? You know, so is it, should we be teaching rubric creation or is it rubric, you know, evaluation and editing?
And I think that's just going to be a continually moving target of like being like, what really are the relevant skills?
As this becomes more prevalent, I also think the potential of setting up students to generate their own content to respond to instead of, you know, the examples I bring into class, that potential of, you know, especially in an online course where you list the readings, but it's always a finite amount of material or finite number of examples for them to respond to. And now the potential for infinite number of examples which they could respond to with all sorts of specificity that I will never think of because, you know, I have my own limitations of my experience, which I'm excited for. You know, I'm excited to see what those changes are and what that looks like in the future.
[00:17:10] Speaker C: It sounds like you are using AI to evaluate lessons and lesson design and plans.
What if, what have you noticed about the quality of the feedback that AI gives back regarding lesson plans?
Good stuff.
[00:17:35] Speaker A: I've had, I've had students evaluate AI generated lesson plans. I have tried the other way. I'm actually for my own practice in the, you know, where I've told students, okay, you have to do something the machine can't do. And then I had the like, oh, shoot, I have to do something the machine can't do when it comes to like giving feedback on assignments because, you know, it does it faster and doesn't want paid. And I like time and, you know, being paid. So like I said, they, they struggle with, and I've even had somewhere and they said, no, that's, that's cheating. I'm saying, like, no, like I'm the one you're giving the assignment to and I'm telling you, like, use this. So like, one example is they had a student who, who was struggling with her writing a bit and said, copy and paste your paper in and, and ask for feedback. Right? Like you, you can do that at home. That's, that's all tools available to you. It's like, no, that feels like cheating. And like, it's one, I, I, I'm the one you're giving it to. Like, it's not, it's not cheating, but I think being more intentional and teaching those practices is where it's really going to head. And to be completely honest, I think there are some skills within our program we're going to have to look at.
Is this still a relevant skill to teach or does this skill now become editing AI generated content? You know, so is it, should we be teaching rubric creation or is it rubric, you know, evaluation and editing?
And I think that's just going to be a continually moving target of like, being like, what really are the relevant skills?
As this becomes more prevalent, I also think the potential of setting up students to generate their own content to respond to instead of, you know, the examples I bring into class that potential of, you know, Especially in an online course where you list the readings. But it's always a finite amount of material or finite number of examples for them to respond to. And now the potential for infinite number of examples which they could respond to with all sorts of specificity that I will never think of because, you know, I have my own limitations of my experience, which I'm excited for. You know, I'm excited to see what those changes are and what that looks like in the future.
[00:20:34] Speaker C: It sounds like you are using AI to evaluate lessons and lesson design and plans.
What have you noticed about the quality of the feedback that AI gives back regarding lesson plans?
[00:20:59] Speaker A: I've had, I've had students evaluate AI generated lesson plans. I have tried the other way. I'm actually for my own practice in the, you know, where I've told students, okay, you have to do something the machine can't do. And then I had the like, oh, shoot, I have to do something the machine can't do when it comes to like giving feedback on assignments because, you know, it does it faster and doesn't want paid. And I like time and, you know, being paid. So I did take some student work and put in there for what does it look like? And understandably, I think the AI feedback on lesson plans in terms of like alignment of objectives, assessments, those sorts of things, it is really not bad. Also does not cut any slack. Like, you know, it doesn't, it doesn't sugarcoat any. Like, I think this could be, you know, it's not bad, but you could do better. Like, no, these don't.
But it doesn't know the student who created it, so it doesn't know the, you know, a lot of my elementary science method students come in scared of science. And so, you know, a.
When I give feedback a content error, I, I deal with pretty gently because they're already nervous about, oh, I don't know enough about this topic. And so it's that, you know, oh, you mentioned this actually blank AI feedback doesn't. It doesn't know that doesn't care about it. So it's just that that's inaccurate.
So I've noticed that that like that to me is the difference in the feed what it does in terms of evaluation.
And I think what as educators, what we still have to offer is okay, but I know the student who created it and can get feedback based on what I know about them.
Not just dehumanize or sort of, kind of objectify sort of feedback.
[00:23:16] Speaker B: It's not trying to teach them anything.
[00:23:19] Speaker A: No, no, it doesn't care about their feelings, Doesn't. Doesn't know what progress they've made.
Just do things line up the way they should.
Does it match the standard? Which is helpful, but I don't know if it's necessarily instructional.
[00:23:39] Speaker B: You had mentioned earlier, you know, and I'm just curious, because differentiation modifications for your classroom context is not just like a fantastic skill that you need to have as a teacher, but a pretty difficult one to make context specific. And I'm wondering whether or not that worked when you tried to modify it with your students. Like, you know, was it a technology modification? Like, did that actually work out when you used it?
[00:24:10] Speaker A: Yeah, so we've, and we've done a few.
In working with some other colleagues, we've done a few lessons where we've, you know, just specifically modeled, you know, okay, we're going to try generating a lesson and then try modifying.
And what we've found to be a better practice is generating a list of 10 possible modifications for the above lesson that address whatever, students with disabilities, students with dyslexia. And then from that list saying, okay, expand. You know, you look at some of those like, yeah, that's not going to work. That's not going to work. Oh, expand on the third option.
And they're not. Not bad, especially when the alternative is a lot of times sitting in a vacuum just trying to think of, okay, what do I do for this student?
And, you know, when, when someone has the knowledge or skills that they go, oh, yeah, this is what I want to try. Great, like, do that.
But as far as the ideas it generates, they're not. Not bad.
And the more detail you put into the prompts, they get better. Now, I do have to share one of our funny examples that when we did this in a class that we went up and we're generating the lesson, we added in some details of, okay, include a fire drill during the day.
And then we had modify it for a student that's homebound.
And what we ended up with worked reasonably well for the homebound student, but it did also have the homebound student during their day going out of their house for the fire drill that the kids in the school were doing.
And which was a nice discussion of, like, hey, most of this isn't bad. Like, you can't copy and paste because you, you know, the homebound kid who's probably there because of medical reasons doesn't need to go out in their yard to practice the fire drill because you had one at school, you know, and, and some of those, like, silly things stand out. But most of it, you're like, this isn't bad.
And.
And certainly when. When educators are dealing with things that they just aren't that are outside of their area of comfort or experience having a. Like, what are 10 ideas of ways to modify this lesson for a student with blank or who blank.
It's a really great starting point, I think.
[00:27:03] Speaker B: Anything else you want to share about your experience using it as a teaching tool?
[00:27:10] Speaker A: I do think one of the most powerful things and using it as a teaching tool or one of the most important things is our vulnerability as educators in showing our students this is a new tool. I don't know how to fully use it. I am trying it, and it may go terribly wrong at some point. And that that's all okay, especially among our education students. So many of them are so scared to make a mistake of any kind.
And being able to model that, you know what? I don't know what's going to happen, and I don't know exactly how to use this, but I want to try. I think it's one of the most powerful things we can do. I also think it does a lot to build rapport in relationships in that, like. And make students feel safe making mistakes when they be like, well, I just watched you, you know, mess this up terribly. Like, I don't feel as bad that I. I'm making mistakes now. And so I think that just modeling, because there, you know, AI will continue to develop and progress, but there's going to be new developments that we can't even imagine or prepare our students for. And so we have to model and teach that. Like, what do you do when you encounter something new? You know, that's. I think that's the relevant skill isn't any particular technology, but just the.
What do we do with things that are changing so quickly, and how do we respond to that, and how do we learn how to use them, you know, effectively, but cautiously and just also just kind of enjoying that process of, like, messing with something, you know, like, there's a lot of fun to be had. And I think sometimes that gets lost when we don't just allow the opportunities to sort of explore and be curious and try a new tool like this.
[00:29:18] Speaker B: I think. I think that's a lesson, like you're saying, beyond just the tool. I hope our faculty can hear that and apply some of that to their own teaching as well. I think having vulnerability with your students is a great skill, and it does definitely build community with your students and relationships.
Well, I just want to kind of recap that it sounds like you're using it as a to generate content specific to your students needs and teaching context, mostly around discussions right now. But you'll probably look at doing a couple different things in the future and we'd love to hear about that.
So thank you so much for joining us today.
[00:30:05] Speaker A: Thank you for having me. Thank you for listening to this episode of FHSU Tilt Talk. Subscribe on Spotify, Amazon, and check us out on the TigerLearn blog or the Tilt social media pages for updates. We'll see you next time.
