Bonni Stachowiak [00:00:00]:

Today on episode number 537 of the Teaching in Higher Ed podcast, teaching effectively with CHAT GPT with Dan Levy and Angela Pérez. Production Credit: Produced by Innovate Learning, Maximizing Human Potential. Welcome to this episode of Teaching in Higher Ed. I’m Bonni Stachowiak, and this is the space where we explore the art and science of being more effective at facilitating learning. We also share ways to improve our productivity approaches, so we can have more peace in our lives and be even more present for our students. I’m excited today to be welcoming back to the show, Dan Levy, and to me welcoming to the show for the first time, Angela Pérez. Dan Levy has been a faculty member at Harvard University for over 20 years, where he’s held various positions related to promoting excellence in teaching and learning. He currently serves as the faculty director of the public leadership credential, the Harvard Kennedy School’s flagship online learning initiative.

Bonni Stachowiak [00:01:21]:

He cofounded teachly, a web application aimed at helping faculty members to teach more effectively and more inclusively. He has won several teaching awards, including the university wide David Piccard award for teaching and mentoring. Dan wrote Teaching Effectively with Chat GPT with Angela Pérez. He also wrote teaching effectively with Zoom, a practical guide to engage your students and help them learn, and maxims for Thinking Analytically, the wisdom of legendary Harvard professor Richard Zekhauser. His teaching was featured in a book called Invisible Learning written by David Franklin. Joining Dan in conversation today and also in collaboration on the book Teaching Effectively with CHAChPT is Angela Pérez. She’s an MPA, an international development graduate from the Harvard Kennedy School.

Bonni Stachowiak [00:02:24]:

Angela currently works as the US partnerships lead at InnovateMATT, an organization that supports school districts to improve math education and serves as a teaching fellow at Harvard University. In the past, Angala has worked as chief of staff to the founder of the African Leadership Group, advised governments on education and job creation profiles in Costa Rica, Lebanon, and Rwanda, and worked as a management consultant. Angela, as you will hear, is passionate about fostering classroom innovations that motivate and enhance student learning. Dan Levy and Angela Pérez, welcome to Teaching and Higher Ed.

Angela Pérez [00:03:15]:

Thank you for having us, Bonni.

Dan Levy [00:03:17]:

Thank you so much, Bonni.

Bonni Stachowiak [00:03:18]:

I I think so often about imagination. We can hinder ourselves by thinking we have to know all of the how to’s before we can really get started. And what I appreciate so much about both of your work is that you really help us take this more abstract thing and get real practical. Let’s do it. So let’s begin with starting pretty simple. Even if I really haven’t gone in and played very much with chat GPT or other large language models, what comes to mind for each of you as you think about here’s a couple, 2, 3 things that would just allow you to start really simple that might have some seemingly small payoff for you. And, Angela, why don’t we start with you? What’s coming to mind for you? Some of the simple things we might just start with and just start playing with it and getting some some benefits.

Angela Pérez [00:04:07]:

Perfect. So the first thing that came to mind for me was the brainstorming phase whenever we are trying to either create a new assignment or, you know, create a new activity for our students. There’s always this moment of trying to face the blank page. Like, how do I overcome that? Right? What ideas come to mind? And I find ChatDBD has been an amazing resource to say, okay. Give me, like, 10 different topics or 10 different ideas that I could that I could use to get inspired about something. And many times, like, I will ask, for example, Chargegbt, give me 10 ideas of topics that I can use to ask a question about bias and updating in our statistics course. And then they would give 10 ideas, and sometimes the 10 ideas are not perfect, are not what we will end up using, but it inspire us, you know, for new topics and, concepts that we can use. I think that’s, like, a very easy, simple thing to get started with LGBT that I would recommend.

Bonni Stachowiak [00:05:00]:

Thank you so much. And, Dan, what’s coming to you as far as the simple things we might get started with and experimenting with?

Dan Levy [00:05:06]:

I think it’s very easy to stay paralyzed, and I don’t know where to start. And so I think the easiest way that you can do is to just pick up something you want to improve in your course. Think about a set of slides that you’re just thinking, you know what? They’re not working that well. Or maybe a class that you have had where you feel like it hasn’t worked that well, and maybe you want ideas for what you could do to introduce more active learning. And then literally simply drag those slides or that class plan or whatever documents you think would provide good context for the AI tool, Chargebee T or whichever one you use it, and start having a conversation as you would with a colleague. And just ask questions of the sort that Angela just pointed. It’s very easy to be overwhelmed right now with AI. If you search, you know, how can I what’s the formula for prompting that I should use? You’ll have thousands of entries come back to you, videos explaining this is the perfect formula.

Dan Levy [00:06:07]:

You know, prompt engineering courses telling you what you should learn or not learn. It really is not that complicated. Just get started, and then you’ll start developing skills. But I think getting started and starting to explore what AI can give you and its limitations is the most important thing, I would say.

Bonni Stachowiak [00:06:28]:

I have had colleagues who aren’t, you know, in this world as much as I have been in the last, couple of years, and and so it it there’s it because it just doesn’t work the way we might predict that it would work. And I just wanna share based on what you said, Dan, that sometimes it could even work differently in the same instance when you’re using it. So I might have created a new chat, and then all of a sudden like, sometimes when I copy and paste a text over, it it remains as part of the body of that thread, and then other times, it creates it as almost an attachment there. And then, I mean, I’ve literally had it where in the same 5 minutes, it’s operating spectacularly well and then terribly, awfully all in the same span. So what I appreciate about what you both have encouraged us to do through your work and and research and writing is to have more of a playful mindset and to back away a little bit from making us think that it’s all about coming up with the perfect prompt, sort of a pass fail, if you will, around a single prompt. And one thing that took me a bit of getting used to I’m sure I’m not all the way there, and I really benefited from getting to read your book. But the idea of it being a back and forth conversation and, Angela, I know that you have a a metaphor or an analogy to help us think a little bit. How how can we make this more into a conversation versus thinking we have to be the perfect prompt engineer?

Angela Pérez [00:07:58]:

Yeah. Bonni, so the way I think about it is that we have to treat Chat2PT as a conversation same as we would with a colleague or with a friend. So take the example of you want to gift something to your partner. Right? Is there is there a birthday and you don’t have any idea? So you start brainstorming with a friend. And then that friend suggests something like, why don’t you buy a spa day for, the 2 of you to go on the weekend? And then you think, oh, that was not a great idea. Or like, I’ve done that before. But you tell them, you know, I like the idea of doing an experience. Let’s use, you know, do you have any other ideas that are similar to a gift, an experienced based gift? And then that friend could suggest other ideas like, what about a concert? What about, you know, going to a theme park? What about a trip? And then it just becomes a back and forth conversation until you reach your final the final good answer to the initial problem that you were trying to solve.

Angela Pérez [00:08:51]:

And the fact that the first answer wasn’t the good answer, wasn’t the final answer, doesn’t mean that that friend wasn’t useful for you in the process of trying to understand what gift you could give to your partner. Right? And so I think I like to I like to think of conversations with ChildGPT as conversations with a human. Right? It’s not just I’m gonna ask you something and I’m gonna explain the perfect answer. But it’s a back and forth process where the other person or LGBT in this case gets to know better what you are looking for until they give you the final answer.

Bonni Stachowiak [00:09:24]:

That is so helpful. And I’m gonna share a quick anecdote, and then I wanna hear Dan’s reaction to this as well to extend our thinking about this back and forth. I was very much looking forward to celebrating one of our seniors. He had been in a couple of classes of mine. He was one of our men’s basketball players, and so I had not been to I kept hearing about this. Oh, it’s the senior night. It’s the senior night. It’s the senior night.

Bonni Stachowiak [00:09:49]:

I didn’t know what that meant. It felt a little bit like it was a cultural celebration I had been invited for, but I didn’t know what the traditions were. I just knew it was a big deal, but I didn’t know, what do you do? Do you bring a gift? And if I were to bring a gift, what kinds of things? So I asked chat gpt, and because I really like I thought, like, maybe it’s like a gift not a gift basket, but, like, you know, a few little small things was what I was thinking. I I’m a big fan of little small things because then you might touch on one thing that somebody really thought was special or something. And so it recommended to me a couple of things that would have been likely to be appropriate, and then it recommended that I buy this young man deodorant and put that into the package for him. It made me laugh so hard because I was just imagining if I had brought, like it just speaking of cultural taboos, that would have been so inappropriate and, so awkward for me. I I mean, obviously, I knew that would have sent something that I should do, but it still made me chuckle. However, in the very same conversation, it recommended to me without me asking it.

Bonni Stachowiak [00:10:51]:

It said, hey. You probably wanna be cautious because there are some rules and regulations around gifts and around college athletics, and we are just in the middle of switching our division that we play in at my university. And so I talked to my friend who’s very involved in athletics, and she said, you wouldn’t have run into those rules this particular senior night, but you will run into those rules this coming this very coming year that we’re about to start. And so it was one of those things where I didn’t know to ask it that information. So it’s telling me information that was helpful. I never would have thought I was just asking about types of gifts and did not know in my prompt that I should include that very helpful advice. Right? And then it also gave me absolutely terrible advice that I didn’t have to ask because I already knew it was really bad. So, anyway, Dan, I’d love to hear some of your reflections on just thinking about what we might learn about thinking of this more like a back and forth conversation.

Dan Levy [00:11:50]:

Yeah. I think back and forth conversation is the right way to approach it. In fact, if I look at the prompts that I write, I often I seem very polite to be talking to a machine because I’m like, please, thank you. And I think part of it, you know, it won’t make much difference on the output that I get, but part of it is putting yourself in the frame of mind that Angela was describing, which is you’re just speaking with another person, and you’re trying through that conversation to arrive to where you are. I I wanna say, one thing about this, which is that, first of all, I believe in human connection, and I believe, conversation with another human being will never replace the emotional pleasure that I get in interacting with another human being. Having said that, I wanna point to a few advantages of using ChargebeeT or any AI tool to have that conversation over, human conversation. The first one is that it has infinite patience. So as Angela pointed out, you can say, give me 5 ideas, give me 10 ideas.

Dan Levy [00:12:56]:

If it would be hard for any of us to come up with so many ideas. So that’s the first one. The second one is that you don’t need all the ideas that it gives you. You can discard as many of them as you want. You’re often in the case of the gift that you were describing, you didn’t need to pay attention to every single, suggestion that it gave you. You could discard as many of them as you wanted, And, frankly, the AI tool won’t get offended. But if you are speaking with another human being and you said, no. This idea is not good.

Dan Levy [00:13:28]:

No. This idea is not good. There comes a moment where that conversation might run into a hold. That won’t happen, if you’re speaking with an AI tool. And then the third thing that I would say that I think is is useful to keep in mind is that particularly for conversations that you have over extended periods of time, I don’t know if this has ever happened to you. Maybe you hire a student to help you with something, and then they produce something, and then they come back 2 weeks later. And then you can’t remember what what is it that you ask, and what was the conversation about. Here, you basically have a record of what’s going on in a way that I think would be much harder to do in our daily life.

Dan Levy [00:14:09]:

So, again, I’m not I’m not suggesting AI should replace our human conversations, but at least for some purposes, I think it actually has some advantages over trying to do this with another person.

Bonni Stachowiak [00:14:21]:

As I was going through and reading the book, I found such such joy in that I could remember not knowing what you were sharing and then just how much faster it would be for someone to get up to speed using this. And that was another thing where now I know that I can not only save conversations, but do the equivalent of pinning them. There’s some search capabilities. And this is a hard thing I find for many people to even if they live this way in their scholarly research, they might understand the importance in their scholarly research of annotating different citations. Perhaps they use some type of a digital references manager of some kind. But then when it comes to just maybe more personal or maybe informal uses, not really seeing a value in something like a digital bookmarking tool, which I go to my digital bookmarking tool more than 10, 20 times a day. But it’s hard because they think, why would I ever do that when I could just Google it? And it’s very hard for me to put into words what you don’t get by just going to a search engine, or in this particular case, what you don’t get if just go to a large language model and that’s where you’ve started on that blank slate, if you will, versus some of the more nuanced, hard to describe benefits. Part of why they’re hard to describe is that we need to be better at our metaphors, but part of why they’re hard to describe is because this is an area that’s changing.

Bonni Stachowiak [00:15:48]:

So just in the last year, many of these models now have what they call a memory, and that memory the first time I remember where it popped up and said, I’ve added that to my memory. I thought, I’m not sure if that’s exactly you know, because, like, sometimes what you’re doing is just playing with it, and you might be, you know, taking on another character. And, the the note taking AI that I was mentioning earlier got confused and thought I was responsible for the Lily Conferences, which would make Todd Socrisyk chuckle because he is the one who runs the Lilly Conferences. So I was like but it just got confused because of the context in which it was looking at the information. It had my bio, but it was also general information about the Lilly Conferences, so that confused it to thinking, ergo, I must be in charge of them. So what else should we be thinking about in terms of either the back and forth that might be helpful or just this idea of saving those conversations and anything that you might wanna add just about some of these large language models, chat gbt, specifically having a memory. What are the implications of that?

Dan Levy [00:16:50]:

So so maybe just very quickly, this is not so much about the memory, but about the, back and forth. I I think it is helpful as in any conversation that the first thing you say is something that sort of sends the conversation in the right way. But I think obsessing too much about what the right prompt is is not very useful. So in the book, we suggest a simple formula for prompting, and you can use anyone, I think, as long as you adopt 1. And this, we adopted from our colleague, Terri Svaranos, that basically tells you, describe the task. What is it that you want Tagigpt or the AI tool to do? Then give instructions how you wanted to do it, and then give context. What is it that, Tagigpt should know about this? Now the context, as you said, is getting a little bit more complex because now TEGPT has a memory or some of our AI tools also have a memory. But I think if you follow something like this, a simple structure like this, it can get you started with experimenting and with seeing, okay, I forgot to mention in the context this thing, and this is why it didn’t do it.

Dan Levy [00:17:59]:

Now you don’t have to write the words task, instructions, and context, but especially if you are beginning, it might be a good heuristic for you to remember. Let me just make sure I include an element of each of this in my initial prompt, and then from then on, let’s just have a conversation.

Bonni Stachowiak [00:18:16]:

We started this conversation saying that your approach really has been, how can we apply this today? And I am excited to announce that we all have the capability to build something that I really doubted myself on. Seemed like it might be a little bit too complex, a little bit too hard, and you have both gifted me with some confidence to experiment beyond what I had possessed before you gave me that gift. So what we’re talking about here, we’re gonna spend some minutes talking about what are called custom chatbots. I would like for one of you to give us just an example of your own or a colleague’s custom chatbot, like, how might one work in a particular discipline or a particular class, and then help build pass it on to the rest of us. How do we can we do this? Is this something that we actually any of us could do? Is it is it harder than we think, or does it actually turn out to be easier than we think?

Angela Pérez [00:19:12]:

I I can go ahead, Bonni. I think well, first of all, I’m really glad to hear that, the book helped you, you know, approach Chativity with more confidence. I think that’s that was a big goal for us to demystify, AI and Chativity for educators and just say, you know, this is something simple that you can start using and integrate it into into your day to day life. And then there are, like, different levels of complexity. Right? And and customized Chativity is something that sounds very advanced and sounds very mystic. But in reality, what it is, it’s just a model of Chat JPT that you can tailor specifically to your needs through specific instructions. Right? So same as in a prompt in a new chat, you would add instructions with, the context and the task. In a customized chatbot, you can also tailor the functionality of the chat GPD model to your needs by adding more context, by adding instructions, and the task that you want that model to fulfill.

Angela Pérez [00:20:16]:

And the benefit of having this customized model versus just a normal char GPT conversation is that you can use this repeatedly over time. Right? You want, for example, to create assignment questions for a specific course, and you know that that’s something that you’re gonna do repeatedly over time doing a course. And you don’t wanna start the chat because if you’re an avid user of Chat GPT, maybe that chat gets lost over time. You could create a customized chatbot that you can then access repeatedly to create specific assignment questions for a course that you’re teaching. We cover some examples in our book, and I wanna share with you specific examples of educators that have used customized chatbot today. And we have, for example, one interesting, customized chatbot from professor Todd Rogers from Harvard Kennedy School. And he uses his chatbot to as a writing editor for effective communication. So that’s, what he specializes on.

Angela Pérez [00:21:12]:

And what the customized chatbot does is its input is normal text that may be long, and then it transforms that text into effective communications using the instructions and principles that Professor Rogers has configured in the chatbot to do. So, like, these are multiple different, customized chatbots that we cover. We also cover personalized AI tutors, for example, that our colleagues, Tedious Varanas and Chartwell did for statistics. We cover negotiator simulations and many others that I’m I’m not gonna, like, continue talking. We could be here forever. But just to say that, you know, we just wanna inspire, different uses that, customized chatbots can be done and that they can be for anybody you know, not only for the most experienced users.

Bonni Stachowiak [00:22:00]:

And, of course, it really does come back to what’s hindering us. But so often, what’s hindering us is just not having quite the imagination for it yet. And then when we get with other people who are also playfully experimenting, as you have done so much with your colleagues, it was very inspiring to see those sparks of imagination that you have used. And then, also, it just sounds like it’s quite infectious there. Dan, what’s coming to your mind in terms of these custom chatbots?

Dan Levy [00:22:26]:

Yeah. So just to be super, super concrete, the example that Angela just used. So suppose you wanna create assignment questions for your course. To create a customized chatbot, what you would do is, as Angela said, you would write the context of your course. What are the goals of your course? Anything that you think Chatgbt would wanna know that if you were starting a conversation from scratch, you would do it. But you would also add any documents from your course that you want Httpd to take into account in crafting those questions, maybe the syllabus, maybe if there’s a handout or slides that you want, to draw from. And then the basic idea is that once you’ve put that structure, you can go to that exact customized, ChatGPT with the instructions and the files that you uploaded and then say, okay. Now I wanna create assignment questions for this topic.

Dan Levy [00:23:21]:

So all you need to do literally is have instructions, which are the same thing as you would do in a prompt for, Child GPT and have the files that you wanted to draw from with the advantage that you don’t have to start from scratch every time. You just go to that customized spot and you do it. So let me just say a couple of quick things and then sort of open it up for more conversation. The first one is that you might think, oh, I don’t know how to do it, but there’s nothing beyond what you have already learned. If you if you know how to prompt Chat GPT, you know how to create a customized bot to the point, that if you wanna see how a prompt looks like for a customized bot, we have a companion site for the book where some of the prompts are there. So you can literally copy and paste and start with one of them like that. So that’s one thing. You don’t need any coding.

Dan Levy [00:24:13]:

You just need to follow the steps to create and then apply everything that you know about how to use, ChatGPT. And then the second thing I would say is that you can think of creating customized chatbots for you. The example that we just discussed is, like, for you. But you can also think about creating customized chatbots for your students. And that is an opportunity that I think many of us could use, particularly as we think about creating some space where students can benefit from using AI in ways that we are conducive to their learning. And so it, in some way, it helps shape their use of AI in ways that are more conducive to learning than it they would be otherwise. Those are a couple of thoughts and reactions on this.

Bonni Stachowiak [00:25:00]:

As you were sharing, I’m gonna give an example that isn’t creating a custom chatbot, but I I promise it does relate to what you just shared. I I enjoy watching videos about different kinds of technology, and, there’s a guy, Mike I forgot his last name. The last half of it is sun, but it’s there’s some other other letters in there to begin with. I will link to this in the show notes so that I don’t entirely confuse all of us. But, he does videos about upgrades to different Microsoft products, including he talks recently about in Microsoft Teams. They have a artificial intelligence bot that can give you feedback on your presenting. And I will admit that when I first saw that, I mean, a lot of my background has been I my my first job out of college was teaching computer classes, and we would go through a train the trainer and learn how to not to say a lot of filler words. And and I remember my manager was very particular about always making sure that you erase that whiteboard so you couldn’t see a single little speck of dust on it when you were done.

Bonni Stachowiak [00:26:04]:

He had such high expectations, and it was absolutely fun to live up to them, you know, that sort sort of thing. I have fond memories. But so at first, I thought like, well, we don’t need a computer to tell us that thing. We just need this wonderful manager that I had at my 22 year old self and things like that. And so many times, like, it’s just so easy for me to forget some of the power differential that no matter how much we might try to reduce it in our teaching, it’s just always going to be present in one form or another. And so I I think of myself as so incredibly approachable. Of course, someone would love to, you know, get some coaching about about their presentation skills from me versus an AI bot. And then I come down to reality, and I had a student once tell me that they were assigned as part of their student government leadership class that they were taking.

Bonni Stachowiak [00:26:51]:

They had to interview a per one of their professors. And she literally said to me something like, well, you were the least worst one. Like, it’s so funny.

Angela Pérez [00:27:00]:

And she she meant it.

Bonni Stachowiak [00:27:02]:

She was very kind, very sweet, sweet young woman, but it she meant that I was the least intimidating, but that didn’t mean that I wasn’t intimidating. So I bring this up because I will say that sometimes I get a little bit prickly. The idea of giving some of what we might bring to our teaching in terms of coaching, giving advice over to a chatbot, or in this case, to an AI tool that is watching the video of you presenting, and it notices that you’re saying, a lot. And in fact, the example he gives on the video is actually on the screen, making this person aware that you’re saying, a lot, that type of a thing, versus when I was 22, it was they would ring one of those hotel bells every time I said, and so when I say, today, I hear myself saying it, and I and I I’m 53 now. We’re talking many decades later. I still have that ingrained in me. So all this to say, what’s coming to your mind in terms of how to have students using chat gpt, but then even whether you’re using it as a coach or you’re using it in some of these other ways that we haven’t talked about yet. So let’s maybe start with students receiving guidance, coaching, input from a chatbot versus a professor, and then we’ll leave a couple minutes for us to explore it from a practical standpoint, students making use of it.

Bonni Stachowiak [00:28:23]:

Does that sound like an okay way to head now? Okay. So who wants to start with, responding to my chatbots taken over coaching guidance that may come from a professor in the past?

Angela Pérez [00:28:35]:

Yeah. But I I loved, your example. I mean, it’s it’s very true. We actually have seen, you know, similar types of uses in in colleagues that we interview for the book. And what I find powerful about those kind of uses is that it makes the feedback from the teacher scalable in a way that would be impossible to do on a one to many basis. Right? So it merges the best of, the teacher’s knowledge and nuanced feedback with the best of technology to be able to give that feedback to all of their, students. Right? And customized chatbots are particularly relevant in this context because if you think about just asking, for example, as a student for feedback on just Chat TPT, it’s very difficult for that feedback to be really relevant to you. Right? Imagine I go in and I record myself in a presentation, but Chat TPT would not know anything about where am I gonna be preparing to be presenting, what’s the level that I should be aspiring to, what kind of audience am I gonna be presenting.

Angela Pérez [00:29:38]:

All those things are relevant to be able to give me feedback that is useful. And if the teacher uses technology and provides technology, the instructions and tailored guidance on what type of feedback is useful, like moving fillers. Right? Like, you know, man gestures, while while presenting tone, etcetera, then it becomes more relevant to the student. Now we’re both Dan and I are not, I think Dan emphasized it, are not fans of completely substituting teacher feedback with, a a technology. We see it as a complement to the teacher’s responsibility. Ultimately, the teacher is the one who and the educator is the one who knows the students best, is the one who is responsible for overall grading and feedback. And giving feedback to students is also an integral part. And as as we’ve seen with many conversations with the students and teachers, an integral part to that relationship that occurs with the students and educators.

Angela Pérez [00:30:42]:

Right? And if we just just delegating that to technology is something that, you know, from our end, we would be wary of doing that we’ve had conversations with the students that have shared that they would see that as something that would be up to the detriment of the relationship with teachers. And so, like, to to to summarize, I think what summarize it is is a very powerful complement, but not a direct substitute of, giving feedback to students.

Bonni Stachowiak [00:31:13]:

So helpful. Dan, what’s coming to your mind?

Dan Levy [00:31:16]:

So just in the example that Angela gave, you can think of the teacher almost being like the designer of the feedback rather than the giver of the feedback. And I think for some skills that required repeated practice where the teacher might not have the time to give that feedback, As Angela said, it could be a it could be a compliment. So public speaking, we obviously, to get good at it, need to practice many, many times. And so the technology could help provide feedback that’s directed, by the instructor. The instructor might have written a book on how to do public speaking or may have, specific suggestions or things that he or she or they want the students to have in mind as they do it. I think it could be a a good complement to do this. Now, obviously, the area of feedback giving, as as Angela mentioned, you know, full of controversy of whether machines should be anywhere nearby. I would just encourage people to think about not as in, oh, human feedback is so much better than the one the computer can give.

Dan Levy [00:32:29]:

That might be true. But to what extent can you give the same level of feedback without the help of technology that you could give, with the help of technology. I mean, one of the principles we have in the book, which we both deeply believe in, is that the key question you should ask yourself when using AI is not whether AI is better than you. Most of the time, you won’t be better than you. It’s whether AI plus you can be better than you. And I think this might be a case where it can in the sense that AI could help you do things faster. It could help you complement you in ways that I think would be very hard to do otherwise.

Bonni Stachowiak [00:33:07]:

Something else that’s coming to my mind is that I I it it surprises me how much this comes up in almost every aspect of my teaching and some of my weaknesses over such a long time. It’s just remembering it’s not over in the sense of just because I may take a risk and introduce an aspect of AI into feedback when candidly, I’m very resistant to it. But I could play with it a little bit. It was once I saw just some of the applications, because it really is context based, and also has to do with what is it that you’re trying to do. And as you said, Dan, if it’s a skill that takes a lot of practice, well, that’s when we really it is we are limited by our human capacity. And so could we introduce the AI? And and so if it gets something wrong, I was thinking about as you were sharing. I know I know it’s not a perfect tool by any stretch of imagination on this show. It’s been criticized quite a lot, but Duolingo, my daughter is taking French on Duolingo, and she’s really into she’s taking it for fun, I mean, for pleasure.

Bonni Stachowiak [00:34:11]:

And yet she says she doesn’t wanna do the talking part, so she just skips past that versus I, dabble. I maybe do it, like, twice a month or something in Spanish, and I will do the talking out loud and get that extra practice pronouncing some Spanish words that are unfamiliar to me. So I I’m probably not gonna be in a position where I can have a Spanish speaker with me at all times to provide that coaching, but I have one in my hand. Is it as good as if I had a Spanish speaker with me at all times? No. It is not. But is it better to practice than to never do it at all? So, I mean, I I I do really like that principle. I’m glad that that came up before we get to the recommendation segment. Let’s spend a couple minutes talking about how students what have we missed? What have we not talked about that you really wanna make sure we share about how students can make use of artificial intelligence?

Angela Pérez [00:34:58]:

I’ll I’ll I’ll come in because I’ve been a student just a year ago, Bonni. So I’ve I’ve I I was, like, wearing to hide while while writing this book, which is, really interesting. And as a as a student, it’s honestly been such a interesting experience to choose ChartiPT. And one of the principles that we also include in the book that it’s very important is that we believe that ChartiPT is not good or bad for learning per se. It just depends on how it’s used. Right? Whether it’s gonna improve or hinder learning. And the most powerful benefits that that we’ve seen, for child gpt is how it helps the students personalize their learning and adapt it to their specific needs. So there’s an example that, we include in the book about 2 students, Sean Norick and Ryan Silber.

Angela Pérez [00:35:48]:

And, you know, the the there was an an assignment in one of Dan’s class that was about learning about risk management strategies. And they were asked to use ChartiPT to learn about risk management strategies before the class. And they take a completely different approach to using ChartiPT to learn about this. So one of them wants a more PhD level explanation because they already come in with a good understanding of risk management strategies. So they want to understand the strategies on a more complex level, while the other one knows nothing about the topic. And so they are asking And you see that ChatDPD’s answers are completely different. And the thing with this is that it would be impossible for an educator to provide both types of explanations at the same time to 2 different students in a classroom. Right? But ChatDpT can do that.

Angela Pérez [00:36:55]:

And, you know, you would you like, we could also ask, is this perfect? No. It’s probably not perfect. But is it better than having a blanket explanation that doesn’t fulfill, perfectly the needs of, these 2 different students? Well, maybe it is. Yeah.

Bonni Stachowiak [00:37:12]:

Yeah. Dan, before we go to the recommendation segment, anything from you?

Dan Levy [00:37:15]:

No. Well well well said. I think I think we have been talking about personalized learning for many, many years, but in my mind, never has it been as easy to help make that happen. And you can make it happen by literally giving students instructions that they could put into Chatibiti or whatever AI tool they use to personalize their learning. Or now you can just create a customized bot that already sort of might start by asking students, what are your interests? What’s your level on this subject? You know, 2 or 3 questions. And then start sort of, a learning experience that’s adapted to that student. Were there tools before to do this? Yes. But they were not as easy to use as this one.

Dan Levy [00:38:03]:

They’re not as versatile as this one. So I think, I I think I think that’s very powerful. Then one more thing I wanna say about this is about customized bots, but about AI more generally. So when we were creating customized bots at our university, one of the things that was very interesting is to hear students’ sort of reasons for wanting to use the bot. And one of the reasons relates to one thing you said before, Bonni, which is that students would say something like, I can ask the bot questions that I would be embarrassed to ask my instructor. So imagine you are a student and you go to your instructor and instructor explains something, and you’re like, oh, I didn’t understand. Maybe you ask once or twice, but then after a little while, even if the instructor is very patient, you might just not feel comfortable. With an AI tool, you know, infinite patience, you can ask as many times as you want.

Dan Levy [00:39:00]:

You can say, I don’t understand. Let’s go in this direction. Let’s go in that direction. Again, is it perfect? No. It it certainly, has issues of accuracy. It certainly has issues of biases, so we don’t wanna dismiss that. But we also wanna think about what’s the counterfactual. What would the student do in the absence of this AI tool? And in some cases, this is better than whatever they would do otherwise.

Dan Levy [00:39:24]:

So not perfect, but as Angela said before, a potentially good complement to you as a teacher.

Bonni Stachowiak [00:39:32]:

This is the time in the show where we each get to share our recommendations, and it may not surprise listeners to learn that the first of my two recommendations is the very book we’ve been talking about. I want to recommend teaching effectively with chat gpt, a practical guide to creating better learning experiences for your students in less time. We’re having this conversation about halfway through the year of 2024. And as of this date, the most downloaded episode ever of Teaching in Higher Ed was featuring Dan Levy, and it was in early 2020, teaching effectively with Zoom. And I got to I was just doing some traveling recently, Dan, and I got to meet so many fans of this podcast, and it was so edifying to hear them share about that. In some cases, they were sharing that that’s right when they started listening. And when they and, also, when they started teaching, like, those two things happened, you know, wait. There’s a podcast about teaching, and I can learn this thing.

Bonni Stachowiak [00:40:30]:

But they were just talking about how they went through such difficult times, but they felt less alone because they had each week these conversations that felt helpful to them and practical. And so it doesn’t surprise me at all that that is, as of today, the most downloaded episode ever, and I suspect that this one is similarly going to get a lot of attention because of both of your approaches to making this accessible to it to us, making you know, helping to build our confidence and making it practical. And so I just really wanna recommend that people check out the book. The second thing speak I know all 3 of us as of this conversation have all been traveling. I really try not to recommend the same thing twice. I think I broke that rule once in 10 years, but will let me off the hook a little bit. But I so I’m gonna cheat a slight little bit here. So I have recommended this app in the past.

Bonni Stachowiak [00:41:21]:

The app is called flighty, the the name of the app. So I’m going to recommend flighty pro, which is the paid version of an otherwise free app so I can get away with recommending it again. Dave and the kids and I were traveling, and when he first told me about flighty, I thought it was just because he’s such a geek about these things. Like, he always loves knowing the kind of plane that he’s flying on, and he really enjoys getting to talk with pilots. Like, he just this is his thing. So when he told me about it, I thought, like, oh, I don’t care what kind of a plane I’m flying. That’s a you know, it’s it’s just not something that I think about very much. And so it’s like a travel app that I never knew that I wanted slash needed because you really can customize it to the level of detail that you might wish to have, but it is so convenient.

Bonni Stachowiak [00:42:07]:

Most of the time, it will pick up on notifications before an airline is even able to communicate with you about really pivotal parts of your travel experience, you know, if you’re gonna be delayed and all the implications of that. It’s also just a gorgeously designed app. I mean, so fun. And then if you wanna share your travel with other people, it’s extremely easy to be able to travel, and then you can kinda see the map of where you are and all the things. So the pro version gives you access to greater tracking, where’s my plane, 4 times faster alerts, calendar sync, sharing, and travel history all in one. So that is my those are my 2 recommendations. And, Angela, I will pass it over to you for whatever you would like to be sharing today.

Angela Pérez [00:42:52]:

I’m definitely gonna be downloading that app. I love tracking all of my trips. I’m become documenting anything that I do. So that was definitely a great recommendation for me. I’m gonna recommend a book that I started recently called The Anxious Generation by Jonathan Haidt, which I think many many people may be already familiar with, but I’ve been following his work for many years. And as someone who is in education and also, you know, believes in the power of technology, but he’s also very afraid of what technology can do to us, You know, I feel I find that Jonathan Haidt’s, thinking really challenges my own ideas and helps me better understand new trends among younger generations. And so I think it’s a it’s a very good book and very relevant in in today’s world for anybody in this field. And I could continue talking about this forever, but I know I’m very short of time and also excited to hear about Dan’s recommendation.

Bonni Stachowiak [00:43:50]:

Yay. Thank you for that segue.

Dan Levy [00:43:53]:

So I’m gonna, do the same as you, Bonni. I’m gonna recommend the book as you and Angela did, and also recommend, technology because, I think, I’m also a little bit of a tech nerd when it comes to using things for improving your daily work. So the book is called Becoming Great Universities. The authors are Dick Light, who’s a colleague of mine here at the Kennedy School, and Allison Jegla. I love the book, and I think the book should be read by anyone, either teaching or, in a role of administrator at any higher education institution. The book is a little bit I know, Bonni, you follow with great interest, work of Jim Lang of small teaching. And the book has that that flavor of, like, look, you don’t need to, you know, invest 1,000,000 of dollars to do some of these changes. And there are so simple and yet actionable ideas for improving what we do at universities that I just found the book tremendously helpful.

Dan Levy [00:44:57]:

On the technology, this is something that Angel, every time comes to my office, chuckles a little bit, but I will recommend anyway. So I have this device called a Stream Deck, which normally, is for streamers. And it’s a device that has a bunch of buttons, and the buttons take you to wherever you want to go in your computer. And so in some way, the devices pushes all my buttons because the pleasure of pressing a button and going to exactly where you want on your computer is just extremely powerful. And just to illustrate with one concrete example how it worked, when Angela and I were writing a book, as you can imagine, we had many documents and websites that we sort of wanted to keep track of. 1 was the book the book manuscript itself. The other one was how we were keeping track of the collaborators and getting authorization from them. The other one was what examples we’re drawing from and so on.

Dan Levy [00:45:54]:

And so I have a computer with 2 screens, and I would press a button in my Stream Deck that said start writing. And then I would see in 7 seconds, those 2 screens divided into 4 quadrants and each of them with document that I needed at that time. So I didn’t have to rearrange windows or anything like that. So doing that produced a sense of joy and pleasure for me that I could start writing whenever I wanted to start writing. I didn’t have to spend all this time getting distracted. So Stream Deck, I have no financial interest in this. You don’t need to be a streamer. It’s very easy to use, but I just found it to be an incredibly powerful tool to, and if you are ever teaching on Zoom or Teams or anything like that, I think it’s extremely helpful in helping you manage the flow of Windows and everything that happens.

Dan Levy [00:46:45]:

So those are my 2 recommendations.

Bonni Stachowiak [00:46:47]:

Oh my gosh. All of these recommendations are bringing me so much joy, and I just wanna echo the satisfaction of pressing one of those Stream Deck buttons. They’re so they are really satisfying buttons to press. I I do need to up my game to get that what our, I know that Dan and I are fans of the Mac power users, what, David Sparks calls contextual computing, but just that idea of now I am writing. I’m gonna have to link to this in the show notes because James Lang had a piece about his writing practice when he gets ready to write what he does, and some of it’s analog. And I remember he he makes a cup of tea and things like that, but how fun to think about the digital practices and the analog practices. Well, I guess this just means y’all have more we could talk about in future episodes as I look at the clock and look at our time. Such joy that you’ve brought me, such curiosity.

Bonni Stachowiak [00:47:36]:

Thank you so much for this book. Thank you for your collaboration and for bringing all of your colleagues and friends into this and catching our spark of imagination into what they’re doing as well. Thank you both so much.

Dan Levy [00:47:49]:

Thank you, Bonni. Thank you very much.

Bonni Stachowiak [00:47:53]:

Thanks to both of you so much for these wonderful recommendations, to Dan for reconnecting on teaching in higher ed and Angela for this initial conversation. So appreciate your time and expertise and really enjoyed teaching effectively with Chat GPT. Today’s episode was produced by me, Bonni Stachowiak. It was edited by the ever talented Andrew Kroeger. Podcast production support was provided by the incredible Sierra Priest. If you’ve been listening for a while and haven’t signed up for the weekly update, head over to teaching in higher ed dot com slash subscribe. You’ll receive the most recent episodes show notes along with some other resources that don’t show up on the usual shows. Thanks so much for listening, and I’ll see you next time on Teaching in Higher Ed.

Teaching in Higher Ed transcripts are created using a combination of an automated transcription service and human beings. This text likely will not represent the precise, word-for-word conversation that was had. The accuracy of the transcripts will vary. The authoritative record of the Teaching in Higher Ed podcasts is contained in the audio file.