– Thanks, everyone. All right, Todd, you ready for the first slide?

– . This is great, let’s begin.

– All right, so yeah, I’m Lew Ludwig, I’m a Professor of Mathematics at Denison University in Granville, Ohio. We’re about five hours southeast of Chicago. So if that helps you out, I think we’ve got folks from all over the world, so glad to be here.

– I’m Todd Zakrajsek. I’m at the School of Medicine, University of North Carolina, Chapel Hill, which is obviously in North Carolina. We’re about the middle of the state. And, also, I didn’t have it on the slide but I’m the Director of the Lilly Conferences on College and University Teaching. So that’s a big thing there, about all. So today we’re gonna talk a little bit about some of the stuff actually I got out of our book. Now, Lew and I have submitted a manuscript as of last week, so the book does take about six months of production time before it’s out, but we’ll pick off a couple of things about it. I just wanted to list some things on here for you. “The Science of Learning Meets AI: The Practical Faculty Guide”, that’s the name of the book. It’s got tons of activities in the book. I’m not gonna read all these because mostly Lew would mute me immediately. So these are some of the activities that are in the book. Go ahead and hit the next slide here, Lew, and we’ll just keep on rolling. These are backwards, so this is great. It was supposed to fill in that way. That’s all right, it’s gonna happen. That’s the way it goes. The other thing we have is a lot of ed psych in the book. The concept of the book is, instead of just saying, “Here’s a lot of stuff that AI does,” or “Here’s how you do AI,” it’s tying in what we know about the science of learning. We already have tons of information about that. So what we’ve done, we’ve took science of learning things that we know like automaticity, backward design, practice retrieval, self-regulation, and we’ve shown how you can use AI to help augment those learning principles. And so that’s the overarching thing for the book.

And now, what we’re gonna do is, for the next slide we’re gonna have three of the big ones we’re after. Part of the book is built on this model, this ACE model which Lew will be talking about in a second, where it’s not just like doing AI, you think about it at different levels. You either adapt something you’re currently doing, you create something new based on that, or you just embed it into the course. And so we’re gonna use this adapt, create, and embed. We picked out three of these, we looked through the book and thought these would be nice. We’ve got TILTing, Mary-Ann Winkelmes’ TILT, the ‘Transparency in Learning and Teaching’, how to tilt an assignment. We have how AI can be used to help you level up on Bloom’s Taxonomy. We picked that ’cause most people have seen Bloom’s Taxonomy before, so we’ll show how that’s done. And then for embedding how you can really pull things into a course there is we picked off backward design, and we picked that one off Wiggins and McTighe’s concept of backward design there, just to help you see a really easy way of pulling it all together. So between TILT, Bloom, and backward design, you also get the three levels of the model. So that’s what we’re gonna do, and we’ve got a short time to do it in so Lew’s gonna take off here and get us rolling.

– All righty. Thanks, Todd. Okay, so hi folks. We’re gonna start off with two many shiny objects and, yes, T-W-O here is intentional. What happens is, as Todd kind of alluded to, there’s just so much talk. We’re coming up on our third anniversary of generative AI, ChatGPT coming out on November 30th, so light three candles or whatever. But there’s just so much stuff going on with AI, and you go to these sessions and you hear all these things, and it’s just like you hear all these shiny objects like, “Try this, do this, do this.” And people are kind of always running around chasing around after these shiny objects. And it’s really kind of hard if you’re new to AI or just trying to get involved with it, especially how it’s gonna help you with your teaching. There’s just too many shiny objects to chase around, so what we’re gonna try to do today is narrow that focus and give you a framework, a model that you can follow, as Todd talked about, this ACE model that we’re building. So that’s one of the shiny objects I want to talk about. By the way, in case you’re interested, there will be minions in this talk. None were harmed in the process of making them, but I am from Denison University, the homeschool of Steve Carell, who was the voice of Gru, so anyway, that’s the shout-out to Steve there. The other shiny object that’s going on, this is the classic “I’ve got a shiny new hammer, and I want to use the thing.” So if you think about it, I don’t know how many of you were around when clickers came out, and people kind of lost their minds, “Oh my God, I wanna do a clicker. I wanna do a clicker in my class.” And they really were like, “Well, why?” “Well, ’cause they’re cool.” “Well, what’s that gonna help with your-” “Well, the students will like it,” right? And it was really kind of confusing for folks why they wanted to use clickers, it was almost like a FOMO, a fear of missing out sort of thing. But what they were doing is letting the technology drive the teaching, drive the outcomes. And that’s not what we want to do. Todd already alluded a little bit to that backward design process. Our main goal, so when I go to workshops and stuff people are like, “Hey, how can I use AI in my class?” I’m like, “Wrong question. What do you want your students to learn in your class? What’s the outcome that you want them to know, that you want them to have? That’s first and foremost. Then you figure out, ‘Hey, how can I assess what my students have done? How do I know if they’re picking up what I’m putting down?’ And then finally, then you worry about the teaching component, ‘How am I actually gonna convey that information?'” We’ll dig into that a little deeper here later on, but we kind of worry about this hammer thing. So I want you to kind of dismiss these two ideas, these shiny objects, we’re gonna try to keep you a little bit more focused. And also focused on our students and their learning, and not just this shiny new hammer.

So as Todd mentioned, we have this model that we call ACE. So Adapt, Create, and Embed. The thing about this model is we’ve developed this in such a way because, again, this is that shiny object thing. So here’s a shiny object and I’m gonna show you how it works in biology, and here’s a shiny object and I’m gonna show you how it works in history. And you’re like sitting in econ you’re something like, “How’s it work for me?” The models that we will present to you today, these examples that we’ll present, are actually kind of agnostic. They will apply to any major or any discipline that you’re doing, and you’ll be able to walk out straight away and use these things in your work. So that’s kind of the nice thing about this model. Again, we’re gonna do this idea of TILT, Bloom’s Taxonomy, and then backward design.

So let’s go ahead and get into it. So adapting, so TILT. As Todd mentioned, this idea is called ‘Transparency in Learning and Teaching.’ Here’s my little startup image. Have you ever had that two- or maybe three-week project that you had your students, you laid everything out that they were supposed to do and you worked, and you scaffolded, and blah, blah, blah. And then they hand something in and you’re like, “What’s this? This is not what I asked for.” And the student are very excited about what they handed in, but somehow there was just this mismatch going on. And it turns out that happens quite a bit. And this idea that Winkelmes came up with in 2009, TILT, can kind of help with that. And what Marianne pointed out was, the thing that we do we’re really good at the task part, we’re really good at explaining to students the task they’re supposed to do. What we forget is the purpose, why are they doing this to begin with? Why are students doing the task? How does it fit into the chapter? How does it fit into the course? How does it fit into the major, even their discipline or even outside in their work life? Why is it important? So how do we get that purpose in there? And then students understand, “Oh, that’s why I’m doing all these tasks.” And then the other third thing she had was this idea of criteria, so these little road marks as students were doing these things they could check to see, “Hey, how is this gonna be graded?” and “Am I meeting these marks that I need to?” So little check marks along the way. And that was the three key elements that she has in TILT. The interesting thing, tons of research. That’s the thing I like about what we’ve been doing with AI here, we’re not reinventing the wheel. A lot of wheels are already out there, they’ve just not been used a whole lot but we’re gonna use AI to pull these wheels out of the shed and use them. That was a new analogy, Todd, write that one down. It turns out that this idea of TILT raises all boats, it helps all students, they’ve done a lot of studies on this. But, in particular, it really moves the needle for first-generation and underrepresented students, so a really targeted audience that we always need to work with. So that’s the nice thing about TILT. Now, how do I use AI to help me TILT an assignment? It’s pretty straightforward. What we’re gonna do is take an existing assignment, we’re gonna upload it to our generative AI, and we’re literally going to say, so anytime today on the slides- again, the slides will be available later on the website, and we’ll also make a PDF to explain all these things, but anytime you see this weird New Courier font, or Courier something, something, Courier font. Whenever you see this Courier font, that’s literally what you can cut and paste into generative AI. So our first prompt’s really quick, right? “Can you TILT – Transparency in Learning and Teaching framework to this assignment?” Now, if you’ve never talked to your AI about TILT before, and you say “TILT the assignment.” It’s gonna make something up. That’s why I put that Transparency in Learning and Teaching in there. Now, the example, let me go ahead, I think I can do this. I’m gonna scoot over to the example real quickly. So the example I have is kind of a straightforward assignment from psychology. So their students are gonna go out in a public setting, they’re gonna identify some concepts, they’re gonna make observation, they’re gonna document it, they’re gonna reflect. To Todd, who’s a psychologist, this makes perfect sense. And he gives this assignment, and two weeks later they come back and he’s like, “This is not what I asked for.” So we need to TILT this thing, we need to TILT this thing. So what we’re gonna do, so that’s my example. I’m gonna go to the TILT example here, or, sorry, the ChatGPT example. And I want to point out, today I’m using just the free model. So this is not the heavy model, the paid-for model, or Claude, or anything like that. This is just the free ChatGPT, just to give you an idea. The reason we’re using the free one, that’s what most of your students have, that’s what you might have access to, so I just want to show you what that would look like. Okay, so here I go. So what I did is I started off by hitting my plus sign, and then I added that file. So that PDF I just showed you, I loaded that up. So that gets loaded up here, so that’s the uploaded file. And then I literally just say, “Hey,” I cut and pasted the prompt in, “Can you TILT this thing?” And here we go, off to the races, it goes through and TILTs. And look at this, the purpose, why are my students doing this? They’re doing this to apply psychology, to strengthen their observations, to reflect on how scientific ideas behave, develop- that’s exactly why they are doing this. It’s like, “Oh, I never thought about it that way.” And then it goes through and does your task, as we mentioned before, that’s the stuff you already had. And then it takes the criteria that you had and kind of makes it into a nice little table. Now, is this 100% perfect? Probably not. You always want to read through this and make sure it’s your tone and it matches, and everything. But, man, this is such a lift that you can do. All of a sudden, I didn’t know about TILT, now I have something that’s gonna get me- TILT 89% of the way on any assignment that I have. And, again, there was nothing magical. I’m a mathematician, I was using a psychology example. There’s nothing magical about the example that I was using. The other cool thing I want to point out, and this is what some of the newer models are doing, especially the paid-for models. It’s kind of a “Yes, and” thing. So it’ll do what you ask, mostly, but then it’ll be like, “Yes, and I’ve got some optional stuff.” So it throws in this optional stuff for you, if you want optional guidance. And then look at this, it even says, “Hey, would you like a student-facing version or a faculty-facing version?” So the generative AI models are getting much stronger, as far as not only in what they can do but then also they’re kind of inferring the next step that you might find useful, and offering those to you. That’s kind of interesting, I think. Okay, that is one done.

So we’re ready for our next one, Create. Create. A Bloom’s level tune-up. Now, I’m assuming a lot of us are probably- Bloom’s somewhere back there. We usually think of it as a triangle or a pyramid. Again, I’m a mathematician. If we do the pyramid version, then remembering gets the most area. And to me that means that’s what students should do the most, and it gets really tiny by the time you get to the creative top. So I like this model by Inaim, where it kind of flips that upside down, it kind of refocuses things. Now, I know what remembering means, I know what understanding and applying- but then I get up in these stratospheres it’s a little confusing for me, it’s like, “How do I know if this is create or evaluate?” So how do I go about doing that? So what we’re going to do is we are going to use Bloom’s taxonomy, and we’re gonna use AI to create something for us that’s gonna hit one of these levels. The example that I’m gonna do is an applying example, but I wanted it to be an analyzing example so I wanna push it a little bit harder, so that’s why I’m gonna have AI help me do that. Okay. So now, be forewarned, this prompt is a little bit long, but I intentionally wanted to do that just to show you what you can do with these things. So first off, I’m briefly gonna describe the context. So here we go with our new prompt, or new font. I’m gonna cut and paste. Here’s the name of my course, and here’s the skill I want them to get. Then I’m gonna give the example. So here’s a typical question, so blah, here’s the example that I want them to do. Now, this is all gonna be one prompt, I’m gonna paste this all in for you in one big shot. Now, the next line- oh, by the way, I’m sorry, here’s my example, let’s do world history. Why not? We’re gonna conduct causation analysis, that sounds very fancy. And the thing is, right now my example is, “Explain how mercantilism shaped Santo Domingo’s trade and plantation using course evidence.” That’s sort of more of an application but I wanna push this a little bit further, and that’s what it’s gonna do. So to push that further, I’m going to create and analyze. So design an assessment that requires students to analyze related to this concept. Now here’s the bonus round, watch this. The task should require students to work with new material, demand reasoning and judgment. I like this next one, be resistant to simple AI completions. Give it a try, right, why not. Include a rubric. And then, finally, provide the task instructions, the scenario or materials needed, and the evaluation criteria, okay. That is a heavy prompt, folks. Let’s take a look at what this thing did for us.

So this is the causation example. Look how long, I mean, it’s a ginormous prompt. So I dropped this whole thing in. Excellent goal, right? It liked it. But now watch. This is actually pretty straightforward, it’s like, “Hey, we’re gonna do this, here’s our task title. Here’s the instructions to students, what we’re gonna do with this thing. Here’s the question that they’re gonna be working with.” And then remember it said to use new sources. Okay, full stop here. This is the free version of generative AI for ChatGPT. It does not have ready access to the web. I can almost guarantee, I would bet Todd’s money that it is making these excerpts up, okay. It just is, okay, just full stop. Just be aware of that. But, again, it gives you kind of an idea of what you might be able to plug in there for it, then it gives the task for the students. There’s the criteria that we wanted, and then it kind of checks through, like, “Hey, here’s how it meets all these goals that you set up for me.” So, again, a pretty heavy-duty prompt but it gave us a very reasonable output. And, again, if you don’t like it you can always push back against it and say, “Hey, fix this, change this. I was thinking this direction.” Have a conversation with it. Don’t think it’s just a static one-and-done thing, you can continue to have a conversation with it. And then this is like, “Hey, do you want a shorter version?” That would be great, right? You know, so again, it’s kind of doing that “yes and” thing for you. Okay. Todd, that is two down. How are we doing on time? Awesome. I’m assuming a nod’s- he has no idea.

– I’m typing the translations over here in the-

– Okay, okay, got it. All right, so next up. oh, sorry, I don’t like this-

– Oh, no.

– Okay, here we go. All right, backward design. Okay, so this is the deal here. So two things are going on. First of all, again, with backward design, a lot of us- I mean, so we are content experts. I’m a mathematician, Todd’s a psychologist, we know our content very, very well. And we always fixate on what we want to teach the students, like these concepts, I gotta cover these ideas, and I might do some neat exercises, but we’re not doing the backward design. That’s the wrong place to start. I should be starting at the end and saying, “Hey, what are the goals I want my students to have after they leave this course? How will I assess those goals?” And then I will think about the teaching. So that’s the backward design. The embed part, what we’re gonna do with this, So again the ACE, so it was Adapt, Create, and Embed. Embed means you’re taking AI, you’re not just adapting something you already have or creating something new. You’re trying to embed something in your course that will happen just kind of as a regular act throughout the semester, and kind of embed both this thing in your course as well as the use of AI. So it just kind of becomes a background thing that you automatically do, that’s the embed part. So what this does, again, the backward design. Outcomes, what your students will know. The evidence, that’s the specific example I’m gonna do the embed for. And then, finally, the learning activities to get them there. And, again, what this does is it helps prevent that idea of the shiny hammer and starting in the wrong spot and worrying about the technology first. So this really focuses on getting your questions in the right order here. Okay, so let’s take a look at this. This prompt is actually fairly short. I kind of use the word unit here. So unit, chapter, whatever the case might be. So I got a learning outcome for this unit, okay. And I think my example for this one’s gonna be an econ example. And I’m gonna say, “All right, it’s gonna take me, I don’t know, three or four weeks to get through this unit.

Now, I want AI to design weekly formative checks that scaffold to the outcome.” So, again, this is that part when we’re doing the backward design where we’re trying to assess whether students are understanding. You don’t wanna wait four weeks until you get to the test and be like, “Oh, they didn’t understand that.” So we’re gonna build this in a week at a time, and we’re gonna scaffold it. Along the way, I want it to describe what are good answers and what are typical misconceptions, so I can be on the lookout for what students might misunderstand. And the thing is, these little check-in points, only gonna take the students about five minutes to do. So that’s my prompt, okay. And, again, the example that I have here is an econ example. Let’s calculate and interpret price, income, and cross-price elasticities, and relate elasticity to revenue. We’re gonna do four weeks to do that, okay. So let me take you out to that one and show you what that guy looks like. So, again, here is my prompt. And perfect, it’s off to the races. So it says week one, here is the- sorry about that. Week one, here is the formative question. And then here’s the solutions that students might give, so the strong examples. Here are some misconceptions that they might have. Week two kind of gradually, not ratchets it up, gives another example. Ooh, we’re starting to get some numbers and calculations here. Week three is ratcheting it up even more, and then week four. Now, as I’m scrolling through here, you see numbers. Okay, a hard pause here again. I’m using the free version of ChatGPT, I can almost guarantee that there are probably some miscalculations in this thing. The free models are just not as good. They’re good at producing words, but they’re not nearly as good as actually doing the calculations. If this was a paid-for model, I would look at it but not nearly as closely as I would for the free model. So the free model, if it creates a solution manual or solution key for you, be very, very suspicious of that and make sure to go through and check it, because oftentimes it does not quite deliver on that with the free models.

But anyway, week four. And then I think, “Look at that, it even gave me a little summary so I can remember what these things.” So it’s going from conceptual to price elasticity, and so on, and how that skill is built. And then it goes, “Hey, would you want me to format this in a different way for you?” So, again, it’s kind of offering different suggestions for us that we can do. Okay, I’m about there, Todd. I’m gonna come around and do a pro tip, and I’m gonna turn it over to you. So I got about two more minutes. Okay, so here’s the deal. We’ve gone through our three things, right, our ACE model, so our adapt, our create, our embed. Again, the idea is that this will apply to any course that you have, it doesn’t have to be a specific major. But how about this, you’ve been out there, you’ve seen some different webinars, you’ve seen some ideas, you’ve got these little different shiny objects running around and you’re like, “How do I convert that into a prompt?” Okay. So for example, you might have seen at a workshop, “Hey, we should rewrite our assignments in the growth mindset language,” right, or, “I want to use AI to create side-by-side samples of needs work, proficient, and excellent, so my students kind of know what they’re shooting for.” Or, “I want to have AI add a five-minute every-other-week check-in prompt for me,” or something like that. So how do I take a simple idea like that and then convert it into a useful prompt that I can actually use in AI? So my ‘Pro tip’ for the day, I’m gonna call this meta prompting. So what we’re gonna do is we’re gonna convert an idea into a useful prompt, and here’s the template for it. Again, this is all cut and paste. So, “I teach a course with this many students,” and then I put the discipline just in case the course name’s not obvious to the discipline. “Help me create a generative AI prompt for this teaching goal,” paste in the example, and then, “Give me a complete ready-to-use prompt I can enter into generative AI, along with instructions on how to use.” So I have a small prompt, I am then going to create a prompt using that, so generative AI is gonna create a prompt for me that I’m gonna feed back into the machine. Meta prompting, okay. Example, “Calculus, 30 students, mathematics. Rewrite my assignments in growth mindset language.”

Okay, last example, here we go. So here’s my short little prompt, meta prompt, right. Now what it’s going to do is it’s gonna say, “Hey, here’s your ready-to-use generative AI prompt. You’re gonna copy and paste everything you see, and then when it says ‘input’, that’s when you’re gonna put in your specific example.” So for me, that would be this swap the wording to the Dweck’s idea of growth mindset. So it says, “Hey, intro calculus, audience, great. Goals, awesome.” Dos and don’ts? Oh my god. Okay, this thing, it went overboard. Oh my god, output format, examples. This is still the same prompt. Okay, the thing went nuts. It constrains text- Oh, there, I’m finally gonna paste the thing in. Good lord, it took forever for me to get to there, and then this kind of output thing. Way too much, way too much. Actually, I was happy it made this example for you, because AI will do that, right. Sometimes it just kind of goes off track. So I’m gonna be like, “You know what, this is a lot. Can you shorten this?” And then here we go. It goes, “Okay,” it says, “All right, you’re gonna have these materials, here’s your rewrite rules, here’s the output, here’s the example. Cut and paste, off you go.” So it actually can create a much shorter prompt for you. So, again, don’t just automatically accept what it gives you. If you’re not happy with it, be sure to go through and try to refine, and things like that, and kind of have that back-and-forth conversation. But there you go, that is our example of meta prompting. Okay. Todd, I think it is back over to you. What do you think, are we done with the slides yet or not?

– Totally counting slides? So, yeah, so we have an opportunity for people to ask questions that they like. I think Lew did a great job with those things, and picked out three things. Again, that first slide showed, or early slide showed there are tons to pick from. So go ahead and type in a chat if you’d like. Giving yourself a second to chat, I’m gonna address one that came in from, is it , for the concept of discouraging students from using generative AI to cheat in their quantitative courses. It’s not just in quantitative courses, it’s the concept of how do you stop students from just using Gen AI all the time. First thing is, I come back to- Lew I think it was you, wasn’t it, that had the students using Gen AI to do their homework and prepping for the exams and stuff. Was that your stuff or was that somebody else I was talking to? Then the exam scores ended up going down.

– Well, so I had formative feedback that I wanted to use on an online thing, and all they had to do was 70% of it right and I would give them full credit. And they were all getting 100% right, and then we went into the written test and it was like, “Yeah, no.” So because they were AI-ing everything, so…

– Yeah, so it’s one of the problems we run into. So in the book we did talk about this a little bit. In terms of doing that, one of the things to keep in mind is just we’re going into a totally different way of teaching. I think if people keep trying to teach the way we’ve taught in the past, then we’re gonna be in all kinds of trouble because AI is changing the game. Doesn’t mean it’s gonna be easy, doesn’t mean it’s impossible here. But, first of all, I’d start out with we have tons of research out there about students cheating in general. I mean, it’s not just AI. What stops a student from getting their homework done by somebody else or going to Khan Academy, two, three years, or actually five years ago, go to Khan Academy, and they’d solve it for you. There’s all kinds of programs out there that would do it as well. If a student had a little bit of resources, they could always get somebody else to do their work for them. What AI has done, has turned it into a very equitable situation and everybody can now afford to cheat at the same level, if that’s what they’re gonna do. But the cheating has always been there. So I think what we have to do is start shifting the focus to how do we teach in a way that doesn’t really benefit them to cheat? One of the things we don’t have time to go into today is we could look at ungrading. Ungrading kind of takes the grading out of it, which was one of the reasons students will cheat, is, “If I’m gonna flunk, I might as well cheat because otherwise I’m gonna get an F anyway.” And so we look at those types of things. Building good community in the classroom. If you’ve got a good community, students are less likely to cheat. The other one is focusing on the learning to whatever extent you can. We like to use the example of going to the gym. I could go and work out at a sports facility, and if I go in there and basically sit down with, I don’t know, a big sandwich and a big drink, and just say, “Hey Lew, could you do 30 sit-ups for me please?” And I can sip away and say, “Now do some pull-ups.” I could have him lift weights, do all kinds of things, use the treadmill and all that. I’m not going to get any healthier because I’m having him do the work. And there’s an old quote from Terry Doyle, “The one who does the work does the learning.” If AI’s doing the work, AI’s doing the learning. And in fact, I was just talking to a friend of mine who’s doing one of the keynotes at one of the conferences we have coming up with a focus on AI. And, basically, she’s from the South so we’re gonna come up with a shirt that says, “If y’all have AI do everything, you’re gonna end up not knowing how to do anything.” And so this idea that if you’re not doing the work you’re not gonna learn, and why take the classes? So if all you really wanna do is get good scores, it’s a different gig. So the point is, I spent too much time on that. It’s a big, big topic, but I think we have to think about how we’re going to be doing that.

– Okay, Todd, there’s a few in the Q&A I can get real quick. So Justin talked about the learning thing. Yes, Justin, I did not know. So I was just going off to the conference and saying, “Hey, TILT this thing,” ’cause my AI had learned about TILT. And then, again, if you’ve never used TILT before, just type, “Hey, what does TILT mean?” And you get like 18 different answers. So once it does understand that you use that a couple of times, mine, it very much understands me, that it has learned what TILT means and puts it in that context. The other question was from David, “The backward design checkpoints are interesting. Your example was based on a unit goal for four weeks. Have you tried a prompt with greater detail regarding the weekly?” Oh my God, you can zero in on this as much as you want, right. You can really, really detail this thing. I will say, as you’re doing this be a little bit careful because, oftentimes- I’m teaching a writing course right now, another whole story. But anyway, it’ll oftentimes, “Hey, I’ll create this handout for you,” and I’ll get a six-page handout. And then I’m like trying to read through the thing, whether it’s any good or not. So sometimes you have to tell the AI, like, “Hey, I want you to brainstorm this with me. Don’t just run off and start doing something but here’s what I’m thinking about, can you brainstorm this with me?” That’s a little keyword that I often use, and then it kind of has this conversation back and forth. So, David, you could probably narrow in on an idea you have, but don’t just keep throwing questions at AI, have it talk to you back and forth and brainstorm that. And you know what? It depends on what you’re teaching. “I want you to be a pedagogical expert in this subject, and we’re gonna talk about this and we’re gonna brainstorm this idea,” right. So it gives it context and helps it brainstorm. I don’t like prompt engineering, that word, it’s like prompt conversation, right. So there you go.

– Yeah, and I like that also with, Lew, and I storm all the time too, as you’re brainstorming and doing it, don’t forget to ask it the other way. As you’re brainstorming and going through things and you feel like you’ve got it, then stop and ask it, “What am I missing here? What might be expected that you don’t see or what is something that a lot of people would put into this material but you don’t see in this one?” So I think that can help. I’m gonna go back up a little bit to Pauline, had a comment of how instructors could disclose to students how AI is used in the design of the course. I think we’re getting to a point, and Lew and I have talked about this too. We should all be disclosing the extent to which we’re using AI. So if you’re in a course and you’ve created a quiz, and for the quiz you’ve used AI to generate the quiz, just tell the students, “I used ChatGPT to generate the questions. I then went back over the questions, I checked them out, I made sure the answers were right. I augmented them, I changed them, we went back and forth. It’s not just a matter of I had it print the questions out.” Just explain what you’ve done. Course design, same type of thing. I could go into a class right now and I could say, “Okay, today we’re gonna talk about classical conditioning. I did go to ChatGPT and I asked ChatGPT what are three things I specifically want to cover today. I then went to Claude and asked for some examples, and then after that I pulled it all back together with the stuff that I’ve been doing for the last 20 years. And here’s what you’re getting.” And by explaining that, what I think it really, really does is it shows the students can use this. I have no problem with students using any kind of AI, as long as, number one, I’m being transparent about how they can use it, And, two, they’re telling me how they use it. And before I let this one go, I’m gonna tell you, it’s not just students with all this stuff, by the way. The conference that I run, I’m now getting proposals from full professors who have all hallucinated references, which means they did not read anything because you can’t read what doesn’t exist. But what they’re doing is just basically saying, “Hey, give me three references, four references that I can tag onto this.” This isn’t a student issue this is a people issue. To what extent do we use this, do we check it, do we admit that we’re doing it? And pulling this all together is gonna be important. So we can’t just hide this stuff, we gots to do it. Boy, I think Lew and I could talk faster but we’d have to have a little more caffeine, which I haven’t had any today so think about that for a minute.

– And, Olivia, I think we’re close on time. Is that true? Maybe went over-

– Close on time. There is one last comment from Michael Fanning, “On the flip side, how can we encourage them at a high level to get over their fear about using AI and plagiarism?” I presume they’re talking about students.

– I thought that had been answered, so I will say-

– Oh, does it? Oh my gosh, sorry.

– No, that’s okay. I thought Lou had touched on it. I mean, maybe he didn’t there too, but I’ll mention this quickly and then Lou can chime in as well. To get over the fear is to do it. I mean, just helping them to do it but give them some assignments, teach them how to do it. I think when we say, “Don’t use AI,” that’s ridiculous ’cause they’re going to. But if we say, “Do use it,” we had the same thing with technology in the past. We can’t assume that students all know how to do this stuff. They know how to do the things they do, but not necessarily the other stuff. So give them some small assignments, help them out to show what it’s doing, show how you’re using it, talk about that a little bit. And the plagiarism part, again, just gotta be really clear about that. And maybe, Lew, you could mention really quickly that scale, was it Watkins that had the scale of usage of AI so that you would know to what extent it was being used? Chapter two, I think it was something like that.

– Yeah. So, again, to Todd’s point, I think having discussions with your students. But the thing is, is that your students, I mean, I’m actually teaching a writing course called “Finding Your Voice in the Age of AI”, and I’m actually teaching a liberal arts meets AI course this semester. And I can tell you, our students do not know how to effectively, productively use AI. They’re just as lost as many other folks are on this, so they are looking to us for guidance and stuff like that. So having these conversations, as Todd alluded to. In the book we talk about, so Ryan Watkins out in the D.C. area, he runs a teaching center and he talked about these ways that, “Here’s an assignment, how could I have-” so you hand the assignment to AI and say, “What are different levels of how students might use AI to help with this?” So just some background information, and so on, blah blah; where it gets to the point, like, if it’s a writing assignment it might actually write the whole assignment. And then you get kind of the scale of like one to five or one to six. Then you as an instructor can kind of take the scale, “I’m comfortable with one through three,” right. And then you can give it to the students, they can honestly rank it as well. Then that gives you some data that you can then have an open conversation of why you’re comfortable with where they’re at or where you’re at, and have that open conversation. I often talk about The Hunger Games, right, and how we’ve got like faculty in District One, students in District Two, and we’re just kind of- If we keep that mentality going and students are cheating, we’re lost. I mean, this is not gonna work out. We have to kind of break down those barriers, look our students in the eyes, have these conversations, help them to realize what we’re trying to do in the classroom and how it’s gonna benefit them. A really, really hard question to ask a freshman right now is the interview question when the person says, “Why should I hire you when AI can do so many things?” Right. How would they answer that? So it’s a tough question to ask, but anyway…

– Hey guys, we are up against time. It was amazing what you got fitted in and I think it was brilliant. And I think we’ll leave it with that kind of end to say, like, we’re all in it together and if we have open and honest conversations about how we’re using AI, I think that’s the best way to move forward. So as I said, guys, this is recorded. It’ll be out to everybody who signed up in about a week’s time. There’s going to be a supporting PDF for it, and watch out for the guys’ book, the book “The Science of Learning Meets AI” is out in the spring, we hope. But watch this space. So, again, thanks a million to Lew and Todd. As ever, very informative and entertaining at the same time. Okay guys, have a good evening. It’s coming up to about quarter to 10 here in Ireland, and the rest of the world, I’m not sure, but in the US it’s what, coming up to about five o’clock, four o’clock for you? Something like that.

– That’s a great thing to end on. Everybody type in what time it is right where you’re at right now.

– Okay. All right guys, thanks very much for your time, and it was very enjoyable. Bye-bye.

– Thank you so much, it was great. Thanks, everybody

– [Lew] Thanks, guys. Fabulous.