Wharton’s Stefano Puntoni on learning with AI, protecting attention, and knowing what you want before you open the chat.
Stefano Puntoni used to give his students a 20-page syllabus. Finding an assignment deadline meant searching through it. Now he makes that information available through a chatbot, so students can find what they need when they need it.
For Puntoni, that is a useful kind of effort to remove. The harder question comes when the effort itself helps someone learn. Working through a difficult problem, forming an idea, or checking an answer may take time precisely because it asks us to build understanding.
In this conversation, the Wharton professor explores how he makes those decisions in his own teaching and writing. He describes using AI to rehearse an assignment, keeping responsibility for his ideas before asking for an edit, and protecting the attention needed to judge the result. The conversation also looks ahead to a university that people return to throughout their working lives.
The Interview
Don: I’m here with Stefano Puntoni from Wharton School. Super excited for you to be here. Thanks for coming. Please introduce yourself, what you teach, what you’re working on, and the world that you inhabit right now.
Stefano: Thank you, Don. It’s great to be here. My name is Stefano Puntoni and I’m a professor of Marketing at Wharton School and the co-director of something called the Wharton Human-AI Research. And my work is on the psychology of human-AI interaction. At the Wharton School, I teach a course called the Psychology of Human-AI Interaction. So that’s basically the theme. And I think it’s such an interesting boundary, this one between technology and psychology right now.
Don: So why don’t we start by exploring the thing you’re most excited about right now? You’re involved in a lot of things, but there’s probably something top of mind that you like thinking about, working on at the moment.
Stefano: Right now, I think one project I’m very excited about is one led by a postdoc at the Wharton School called Ben Lira that looks at the impact on intergroup relations of digital twins and LLM-instantiated synthetic humans.
Basically, what we do is get people to interact with an out-group member and see whether they learn something about that population that they don’t know much about, and maybe they often stereotype and perhaps dislike, by talking to a synthetic version of that out-group.
For example, we find that Democrats and Republicans tend to like Republicans and Democrats more after they’ve had an opportunity to spend a few minutes talking to a bot impersonating a member of that out-group.
Don: That’s super interesting. And it’s fighting against the stereotype that we create for people. But I wonder, a digital twin, is that also a stereotype of what the person they’re representing is?
Stefano: In fact, it is. And we do see that in the data too, but it still helps. And in this case, we were looking at AI-based interventions that can help reduce political polarization. But you can imagine also commercial applications in the domain of marketing, for example.
I’m a marketing professor. I find it interesting to know how decision-makers in companies can learn about customers. And maybe one way that is opening up right now to learn about customers might be not to talk to customers, but to talk to computers, which is a weird idea. And we don’t know exactly how to do it and how well it works and when it’s appropriate. But it seems an interesting avenue, and I think there is some early evidence that there is value in doing that.
Don: And I think this is creating a little bit of space to start asking your potential customer more questions, right? Versus, “I know about them. They generally feel this. They want this.” Just to have a few questions, actually, and start engaging your mind in a different way probably helps, probably moves the conversation.
Stefano: And I think that is maybe even a special case of a much broader application of AI, which is to do a better job. Whether you are a salesperson—I know that many people in sales now would practice or do some kind of extra dry run with a chatbot impersonating a customer to understand how to pitch things.
The same for teaching. Now, many teachers do that; I certainly do. Whenever you have an assignment or you develop some new ideas, you want to know: Is it too complicated? Is it fluent? Is it working? Obviously, you want to test on people. But for the first cut, I find it quite useful to have a student bot instantiated through ChatGPT, where you can figure out some of the mistakes you make. And typically, we always make mistakes.
But that’s probably true in a lot of other walks of life, where you can maybe pilot or try things out before you encounter a real audience.
Don: I love that idea. Because even when you’re preparing a presentation, or you’re doing a keynote or something else, it helps to put the words in your mouth and get them out without even…
Stefano: Do you do that?
Don: Of course. Because it’s one thing thinking a thing, and it’s another thing saying a thing. And then the next level is saying a thing and having a reaction back. That is this progression. And this is perfect, takes us into the space of learning.
I was just at our AI Skills Jam for K12 Teachers. We do these enablement days of bringing teachers together and teaching them about the tools. And I met a new teacher, two years in, a middle school teacher. And the first years of teaching are hard, right? No matter what, whether it’s K–12 or higher education, it’s hard. There’s a lot you’re carrying. You’re building materials. You’re interacting with the class. You’re holding them.
And he just talked about testing ideas out, the same thing. He didn’t create a digital twin. But he said, “I test these ideas out. I get feedback. Is this too complicated? Will this support the range of students I have?” And as a second-year teacher, he said, “I feel more confident. I feel more prepared. I feel better situated to really be a teacher to my students.” And I’m curious how that fits in.
Stefano: Nice example. And I think that in the last six to eight months, I’ve seen a leap in some of the capabilities of these models that now can enable you to do things which maybe a year ago you wouldn’t be able to do. Certainly, I’m thinking about my academic work, the way that I now use it as a proofreader for my articles, or whenever I have it pretend to be a mock reviewer. That’s another audience that, as an academic, you care a lot about: the academic reviewers of your articles.
And ChatGPT Sol, for example, is very good. So it’s been very helpful in making sure that the work is a lot more polished and ready, in a way, than maybe it would have been before.
Don: And there’s also this weird line of, at what point does it feel like cheating? Am I too prepared? Do I have these amazing tools? Is it okay to do that?
Stefano: I think it’s a good point. One way I think about that is to say, every job has frictions. And some frictions are good, and some frictions are bad. And you want to get rid of the bad frictions. And you don’t want to make the mistake of making it easy for yourself in the ones that are good, that you should really put the effort in.
And so maybe that’s valuable advice for everybody to think about in your own line of activity, whether that is professional life or personal life: In this context, what frictions are good, meaning they are needed for the kind of outcomes that they care to obtain, versus bad frictions?
These are things that I’ve got to do, but they are chores. They are maybe things that actually are not core in either achieving the outcome or building my capabilities or signaling to anyone we care about what it is that we’re doing.
And so differentiating the two can sometimes be hard. So I certainly see that in teaching and learning. For students, there are certain frictions that are bad. So we want students not to have to struggle with certain elements of the learning process. And the easier we can make it, the better it is.
Don: And can you double-click on that? I want to spend a little bit of time here, because where is friction good, where is friction not? That’s arbitrary. There’s probably a great line.
Stefano: And you’re right. The difficulty is that maybe my bad friction is your good friction. So that’s one challenge. I think, to some extent, it is, in many situations, a personal choice to decide what parts of an activity are meaningful and worthwhile, and which ones maybe are not. And in many situations, it doesn’t have to be a right or wrong solution, just a different taste.
But probably in some situations, there are objective answers to that question, and learning might be one of them. So if I give students a thick, 20-page syllabus—believe it or not, I used to be doing that—and they have to find information about the deadline for assignment number five on week number seven, that’s bad friction.
It’s just bureaucratic nonsense in some ways, but of course, it’s part of the process. But now I can feed that to a chatbot, which I do for the students, and then they have the service that they can access to make sure that they get information that they need at the time they need it in the smoothest possible way. Now, that’s an example of a bad friction we can get rid of because of AI.
Now, the problem is also that good friction. So maybe you give them some homework, or you give them a take-home assignment. And now the tendency, the desire, is, “I want to hang out with my friends, and we’re spending all this time doing this complicated stuff. So let me get some help.” And even though you think you are being diligent because you’re checking the answers—just copy and pasting—the truth is that by removing that friction, you are making it harder for you to build some kind of deeper understanding of the problem. And so it does hurt learning.
So I think one challenge that we all have as educators, as parents, and as students, too, is, over the next few years, as the capabilities of our AI systems are exploding, to figure out how we navigate that decision about bad frictions and good frictions.
Don. I feel like this is such a core conversation. And that leads me to: What are the skills we should be building? I think that is the next step: the good frictions to build muscle, the bad frictions to get rid of the bureaucracy.
But ultimately, the good frictions are building skills and muscles for something. What are those things that we should also look at objectively, with that framework? Because we can go arbitrarily, or with the ones that feel good and ones that don’t. But there’s probably some bigger framework that we could go to for that.
Stefano: Yeah. So I was making the distinction between good friction and bad friction, and you were making this muscle analogy. You want the good friction because you want to work out. Then the next question becomes, which muscle do you want to build?
Don: That’s exactly right. So what are the muscles?
Stefano: I don’t know that there is a one-size-fits-all. There are probably directions you can point to. That’s something I try to do in my teaching: make students reflect on that “which muscle” question.
You can imagine a few. Some have to do with the value that human labor brings that might be distinct in a particular environment. There are some obvious cases. For example, people are always going to want to go and watch a basketball game. Probably not if robots are playing. There might be something about the human endeavor, competition, excellence, beauty that we appreciate in human skill and achievement, which can never be replicated by anything else. That’s one big domain.
But there is more. Certainly leadership. I think algorithms can manage, but can they lead? That requires role modeling, personal connection, motivation, inspiration, and ultimately, it may require human communication and persuasion, feeling personal connection and trust.
Beyond those softer, interpersonal skills, which I think are on everybody’s radar—we want to be as good as possible at doing those things nowadays—I think there may also be some things in the domain of cognitive capabilities and know-how that one can point to.
One of them is pretty obvious: proficiency at using these tools. I think most jobs in the economy will be touched by AI in some shape or form. In some jobs, it will be a small part; in some jobs, it will be a massive transformation. But wherever you lie on that continuum, it seems unlikely that many people can be successful in their line of occupation or related ones unless they learn to adapt and become proficient at using that tool.
There are basically two ways of thinking about that. One is to say, “I’m going to upskill myself through learning how to do this, be good at Codex or whatever.” We call that coping strategy “direct resolution,” where you have a threat to your competencies and maybe autonomy, and you are going to deal with that directly by saying, “I’m going to make myself the best possible complement to that technology.”
Another one we call “fluid compensation.” In that approach, what you’re doing is saying, “This is a part of a job that is going to be increasingly, for a variety of reasons, done through machines. Let me pivot and be more in this part of the job where machines are less likely to have much of a footprint.” Maybe because we don’t let them, even legally, or because the capabilities are not quite there. Maybe they will never be there. Who knows? And then say, “I am going to redirect.”
Imagine that I’m a content marketer. My job is to write content for a B2B company. Before, a lot of your time was spent writing the content. Now, increasingly, you might lean on AI agents to help you build the content. But the next question becomes, “What content is it that we need? What strategy do we put around that?” Maybe you pivot to the strategy piece.
This is an example of two different ways of thinking about upskilling and reskilling. I think probably most people are going to fit. Maybe for many of us, it’s going to be a combination of the two.
Don: If we want to survive. I feel like it’s a moving target, though, constantly moving.
Stefano: There’s a whole academic graveyard with papers that come out in preprint saying AI cannot do this. Then they never get published because six months later, they can.
Don. In the education space and in the work space, I imagine more than ever there’s going to be a constant need to pivot, to build new skills, to adapt to the new tools that are out there. If you’re doing the thing that the tool can do, that’s not helpful.
Stefano: That’s another muscle people need to build. It’s a bit of a tricky one, a squishy one, which is the ability to change.
I think change is hard for everybody. But it’s hard to imagine being successful in a world where technology is changing so rapidly unless you not only accept that change is happening—you cannot wish it away—but also embrace it and say, “I embrace this uncertainty because uncertainty brings opportunity, and it brings excitement, and maybe also some fun.”
Fundamentally, I don’t quite know how to teach people to accept change and embrace change.
Don: Especially when there’s not a choice either, right? I feel like this is now being imposed. These changes are just there and everywhere. It’s reactive, or you’re somebody who is looking for that adventure anyway, and that’s who you are.
Stefano: In that sense, I think the longer you’ve been doing something, the more difficult it is to adapt. There are a lot of conversations around jobs. Most of the concerns have been around entry-level positions.
You talk to college students all the time. You probably hear a lot of this about entry-level jobs and internships and things like that. Personally, I’m actually worrying more about the older people, the people my age, because I think for them to adapt might require building that muscle. With young people, they are natives. I see what our students do.
Stefano: It’s just amazing.
Don: If you’re older in the workforce, you build this experience and expertise, and it is a huge foundation. It reminds me: I was in Italy, as I mentioned, a couple of months ago, talking to a lot of professors. One of the conversations came up about a professor saying, “I have so much experience. I’ve decades of experience in this field. I’ve built great muscle, great experience. Actually, using AI is power.”
This is the opposite side of this argument: using AI is really helpful because I also have this judgment, experience to relate, to connect, to evaluate the responses AI is giving me.
But students coming in are fresh. They don’t have that depth. What are the skills, what are the muscles? They don’t have that judgment to bring in or that comparison to bring in.
Stefano: I think that’s a good point. One of the most interesting frontiers in behavioral science and social science around AI today is that question: if you have a population that is heterogeneous, variable in some traits and capabilities, and now you throw AI into the mix, what is that doing to the dispersion of the outcomes across that diverse population?
There could be two different results. The first is that the ones who were performing less well can catch up to the performance of the other ones. It’s basically reducing inequality of outcomes within that population. Now you have the entry-level person who can deliver reports at the level of a very experienced consultant or whatever.
On the other hand, you could have the opposite pattern, where to be really proficient at using that technology requires a lot of expertise. Those who do have that expertise now can run away.
It actually is a multiplier of your effectiveness. Then what we see in that situation is that inequality of outcome within the population will increase instead of decreasing.
In the literature, you can see both patterns, depending on the task, depending on the population. I don’t think we have a good grasp, as of today, of what factors make it go one way or the other. But understanding those factors is going to be very important for bringing that technology to the workforce. In every role, you’re going to have some kind of impact, and you want to understand and anticipate whether that is going to be increasing inequality or decreasing inequality. We don’t know yet.
Don: I think there’s that pressure. You have these tools to perform better, to create more polished products, to do things faster, to do more. There’s that pressure. I think the workplace understands that, and the expectations are higher. But if you don’t have that depth of experience, you’re bumping around the surface a little bit and relying more, potentially, on tools than someone who has a lot of experience and can do this hybrid thing, because they can bring a lot more to the table.
Stefano: I think that is taking two shapes. The first is in initialization, and the second is verification. Whenever experts are working with AI, the way I think about it is like a sandwich model, where you have the human at the beginning, the human at the end, and you’ve got a lot of AI in the middle. Sometimes the sandwich has a lot of bread and sometimes a lot of condiments. But basically, you have a combination of the two.
It’s always human and AI. It’s unusual that it is only AI or only human in knowledge work nowadays. To do the verification work, but also to do the initialization work, you’ve got to know where you’re aiming. You have to be able to judge quality. When the models are so advanced and the quality of the output is so hard to judge, it actually requires a lot of expertise to be able to say, “This is good,” or “This is bad.”
Don: And really careful scrutiny, too. These things don’t just pop up; they aren’t obvious. You have to have that lens of expertise.
Stefano: Which brings us to the other muscle that we haven’t talked about so much: the ability to focus attention. I think that is a problem. Everybody talks about the human in the loop, but there’s no use having the human in the loop if the human is asleep. You need to be sharp, and verification work is very hard work.
If you’re a coder, half of the time that you’re coding, you’re really going off memories, almost copy and paste. You daydream a little bit. There’s obviously knowledge work, but there are moments in which it is, “Okay, I need to initialize this routine or whatever.” It’s easy. With verification, you don’t really have that kind of downtime. It’s all full on. You do four hours of that, you’re spent. It’s very hard work.
I think we’re losing attention really at the moment when we need it the most, because work is becoming more intense. We are losing the capability.
Don: That’s good. We’ve been losing it for years, too. It’s the pace of the world and all the stimuli that are coming in. Those things have reduced, I think, attention span. But I imagine that’s out there. Now, like you said, when we need it most, we need to build that muscle again.
What do we do? I think about the skills we’re building for learning, and how do we help people use these tools? What are some of the approaches to do that? What’s that balance?
Stefano: Put the phone away. That would be a good one. And not just in general to cultivate attention. I advise our students to immerse themselves in long-form media, whether that is a thick book or a five-hour podcast. But do something you cannot consume in five seconds.
And learn to be patient and learn to be a little bored.
I think that’s one. Then we talked about the other ones. How do you nurture this acceptance of change so that it becomes an opportunity, not a threat? How do you build the right type of skills that make you a complement to AI technology? You don’t want to be a substitute for AI technology, because, in fact, AI technology is going to be a substitute for you. You want to be a complement. It should be human and AI together. How would you bring that multiplier?
Then that ability to inspire, connect with others, communicate effectively, be trusted. All those softer things we were talking about at the beginning.
Almost like an emerging framework, I think.
Don: That’s one of the things I think about. What’s that framework? What are those things that we hold true and are careful with? What are the other things that are more flexible?
I wonder, in your world, as an educator, as a researcher, where do you feel yourself leaning as you have this growing companionship? I see this as a partnership. More and more, I think about the partnership with AI. What does that partnership look like? What is the part you need?
If you think about the range of jobs somebody can do, like you said, it’s not a hard line. If I’m not a writer, and that’s not really part of my job, I might lean on AI to do my writing for me. Whereas if I’m a writer, that’s the muscle I want to build. It depends on who you are and what you do.
For you, what does that partnership look like now? How is it growing? Do you see a direction?
Stefano: I think I struggle like everybody else to negotiate that bad friction, good friction issue at the beginning. I sometimes find myself using AI in ways that I’m thinking, “If I do this a thousand more times, I won’t remember how to do this thing. And I maybe should.” Other times, there are skills I’m very happy to lose. Honestly, there are lots of things that, say, my parents’ generation knew how to do, which I’m perfectly happy not knowing how to do.
Don: Yes. How about remembering a phone number?
Stefano: Lots of stuff like that.
Don: Maybe that’s important.
Stefano: In a way, we are a little dumber in that way, but does it matter? Actually, no. We’re smart in some other ways. So I don’t worry too much about that. But, for example, I find that if I’m approaching a blank page and I have a chat with me to help me, I basically don’t think as much. So I made myself the promise that I will not write with AI; I will edit with AI. The edit could be very extensive. It could be actually writing most of the words.
But I need to know before I talk to chat: what is it that I want to get out of this? And what are the ideas? Because if I don’t do that, it’s too easy to find myself just leaning on it, and then I become a copy-paste machine, which is not the right thing.
Don: And AI can do that too. AI can copy-paste for sure. I’m having the same experience where, at the beginning, I was using AI a lot, and I was just building a lot of stuff. I was like, “You go do it, go do it.” I had an idea, maybe, but then it could take me in a left direction. I feel like I wasn’t paying attention.
Now, I literally will not sit down in front of my computer until I have an idea of: what is the work I need to do? What is most important?
Stefano: So what do you do? Do you actually draw on a piece of paper?
You do that? Like a diagram on a piece of paper?
Don: I do. I have big pads of paper. When it’s a more extensive thing, or I need to think things through, or I need to get all the stuff somewhere outside of my brain, I’ll just get a big drawing pad and start sketching stuff out.
Stefano: I actually do that too.
Don: Otherwise, it’s very easy to be taken for a ride. And so, coming in with intention.
Stefano: There will be some tasks where that is fine. It’s literally something you don’t want to do.
And then you are happy to outsource this to an AI system. But there will be things that you do want to do.
Don: And not losing your intention and purpose. I think that’s really important. And I think that maybe is the muscle too. What do you care about? What matters? What are you trying to do? What are you bringing to the table? Because you’ve got this superpower, you’ve got this whole army of power on your side, who’s ready to do the work for you.
Stefano: I like your use of the word intention. I think the biggest fear I have is that I end up wanting what the AI wants me to want. And I think you never want to let an AI system make those decisions for you. That’s something you want to keep almost at a moral level. It seems like something that we want to actually, if anything, do more of and say: what is right? What is wrong? What is appropriate? What is inappropriate?
That kind of judgment, you should get as good as possible at doing, because there’s so much more you can do with AI. I think having the decision, the good muscle, the moral muscle. And say, “Okay, if AI is an expansion of my capabilities, then there’s so much more I can do. And there’s also much more I should not do.”
Don: Because just because you can do it doesn’t mean you should. I like that reminder of AI being this amplifier. And if you’re not bringing anything to the table, it’s not amplifying anything. It’s just doing it. It’s like I dropped this seed, or speck of dirt. It will amplify that in a way that you didn’t intend. And so coming in with that experience, with intention, with purpose, is super important.
I want to shift gears a tiny bit, and we’ll probably come back to this. But I think about schools, the classroom, and K–12 and higher ed, and definitely K–12, where I think it matters across the board, whatever grade. Relationships and psychology are as important as subject matter expertise and everything else. You ask a K–12 teacher what the most important thing is, and they will say relationships. And you can’t replace that.
But it’s a very emotional time, a very psychological time. And AI, I think, there’s all sorts of reactions to this thing as coming either to take our jobs, or our purpose, or what we value, what’s important. Why is it so emotional? I feel like this is a space that you’ve probably looked at.
Stefano: We’ve never had a technology like this. We’ve never had a technology that can do things that are so humanlike. And because of that, it’s a mirror it’s holding to us. And so it’s a challenge in that sense.
Just to illustrate what I mean by this, if you went back maybe a decade or two ago and you were reading about human exceptionalism—what is different with humans?—people would talk about their cognitive capabilities, their prefrontal cortex thinking abilities. You do that now, and you don’t hear that story. You hear a different one, which is more about our emotions.
What technology has done is to change the reference for what is not human. And when you think about what makes humans special 20 years ago, you’re thinking about other animals, other living species. And when you think about what makes humans special today, you think about AI, because AI has become so humanlike.
And I think it’s a momentous change. I do believe very few examples of technology in the history of our species have that kind of depth of impact. And I think it’s so exciting to be alive today.
Don: And the crazy thing is, things are changing so quickly that we don’t know what next year is going to look like. We don’t know what a couple of years are going to look like.
Stefano: But what I do think is that AI is going to get better. In a way, AI then becomes an opportunity for us to actually pay more attention to those human things: emotions, well-being and connectedness, and the moral side of our lives, all those kinds of things. People are worrying about jobs, they’re worrying about instability. I think, ultimately, my hope is that this is just an opportunity to become more human.
Don: I love that. We can’t see the future, but how do we spend our time now? And in K–12 and in schools, I think that connection with students, you’re never going to replace that. And again, subtly guiding student learning, there’s an art to that. That is not just a logistical path that a system can build. There’s a real art.
Stefano: Yeah. In K–12 education, an inspiring teacher has a lifelong impact on a lot of people. It’s amazing what impact a person can have.
Don: These change the direction of the course of people’s lives. You ask most people, “How did you get here?” And often it’s a teacher. It’s one teacher, it’s a series of teachers, but it just takes one to literally change the course of your life. I don’t think that’s ever going to go away. We need to lean in on building those spaces where that happens.
And how about in higher education? What does the future look like to you in higher ed? I know that’s a hard one.
Stefano: I think things are in flux. There are a lot of big changes. Some of them have to do with technology, but there are also other changes. And so the sector, I think, faces some big challenges that have to do with funding, that have to do with globalization, that have to do with regulation.
And then you have technology that is disrupting the traditional on-ramp to professional classes, which our students used to take. So we have to think about how to support students. So many of them now try out startups and so on. Can we support them in those kinds of endeavors more than maybe we ever got used to?
Then you have the pedagogy itself in the classroom. How do we use or not use this technology? Again, bad friction, good friction, very hard to do. Assessment is a nightmare right now. I’m bringing up the paper-and-pencil exam.
There’s a lot of hard decisions like that. And then even the bigger picture, I think, long run, is really about having to shift away from this pipeline model that we’ve had for a very long time. Basically, where we take people at age 18, we teach them entertain them, mind them for a few years, and then off they go at 21, and we never see them again, to a world where, because of change—and hopefully this tends to embrace change—people will need to retrain and upskill and discover new passions and interests throughout their life.
And so this pipeline model: come in as a teenager, leave as an early adult, and then the university hopes that they maybe remember us and send us some money. I think we need to think about more flexible models, maybe executive education and certifications and other ways in which we can keep touching those lives and helping people throughout their careers. It’s not just a long goodbye, but it’s more like a tool. Universities are there to support you throughout your career.
Don: That’s amazing. And that’s a vastly different model, right? It is leaning into lifelong learning, really, and universities being actually a part of that whole journey, not just a moment in time and life and learning happens elsewhere. But there is potential for universities to be.
Stefano: Lots of questions. Who’s going to be doing that? Are you going to have specialized institutions that are going to serve that need, that are going to complement what traditional institutions do in the degree programs you’ve grown used to over many generations? Or the same institutions, because they have the infrastructure, the know-how, the connections, that actually are going to venture more into that lifelong learning, as you call it?
Don: And probably good for universities to do that too. That sounds like a really interesting model. Last question for you: what are you most excited about? Looking forward today, in the near future, what are you most…
Stefano: My course is in the third quarter of the academic year, so it starts right after Christmas. And my rhythm is that I start around now thinking about the next run, because I teach an AI course. Basically, every year is a new course.
And so I have now three months where I have this huge library that I collected in the last few months, and I keep collecting new academic insights, new reports, some nice examples. And I’m going to slowly craft those into a narrative that I’m going to deliver to the students in January. I find it’s a lot of work, but I think it’s extremely rewarding, an interesting thing to do. So I’m excited about that.
Don: Well, maybe I’ll stop by in January and sit in a little bit!
Thank you so much for joining today. This has been fascinating. I feel like we’ve just started the conversation versus covered it all, so maybe we talk again sometime soon.
