Explore

Class Disrupted Tackles AI: Exploring its Application for Teaching and Learning

Launching their mini-series on artificial intelligence, Michael & Diane dive into where the tech is useful in education — and where it’s disruptive.

Get stories like this delivered straight to your inbox. Sign up for The 74 Newsletter

Class Disrupted is an education podcast featuring author Michael Horn and Futre’s Diane Tavenner in conversation with educators, school leaders, students and other members of school communities as they investigate the challenges facing the education system in the aftermath of the pandemic — and where we should go from here. Find every episode by bookmarking our Class Disrupted page or subscribing on Apple Podcasts, Google Play or Stitcher.

At the outset of an AI-themed season, our hosts take stock of their prior assumptions, hopes, and concerns about the technology’s applications in education. They dive into where they see it being used to make adjustments to the current educational model and envision how it could be applied to revolutionize learning. 

Listen to the episode below. A full transcript follows.

Diane Tavenner: Hey, Michael.

Michael Horn: Hey, Diane. Good to see you.

Diane Tavenner: You too. I spent the weekend on a tradition I think we have talked about before, which is we hold a holiday party every year for what are now old friends. Because I think this is our 27th annual, if you.

Michael Horn: 27th annual. Wow.

Diane Tavenner: Yeah. And it just, it makes me appreciate longevity and just like I have such gratitude for deep, long relationships that have built over time. And yeah, it’s just really, it’s a good fill-me-up for the moment.

Diane and Michael’s AI Priors

Michael Horn: Yeah. That’s amazing. We’re obviously recording this as we approach the holiday season, if people can’t figure that out from that intro. That’s an amazing place to start. And the gratitude you have around that, Diane. So very, very, very neat. Let’s lay out what we’re doing for folks today.  And as we get into a little series on the topic that we talked about in the first episode back, which is artificial intelligence. You want to lay it out, Diane, what we’re thinking?

Diane Tavenner: Yeah, I think as folks know, like, we are now following our curiosity and we’ve been doing that for a while. And, you know, I don’t think either of us are just like 100% all in on AI, like huge evangelists. And I do think that we’re at a minimum, cautiously optimistic about the possibilities of it. And so we’re just curious about it. And I think we find ourselves kind of talking about it and asking about it. And so we are going to do a little exploration. We’re not exactly sure. We’ve got some ideas of the format and whatnot. We’re not exactly sure how long it will last for, but we thought we’d just kick off today with where we’re starting that exploration. And I think I, personally, I think you’re with me. I hope I end in a different place, quite frankly, I hope I end in a place where I’ve, like, learned some stuff and talked to interesting people and, you know, maybe think a little bit differently. Hopefully smarter than I am now. But today we wanted to just kind of lay a foundation of where we’re coming from based on what we know so far.

Michael Horn: Yeah, love, love that intro. And what I would add is it’s obviously a hot topic in education. Everyone knows that. But I think what’s also interesting to me anyway has been how OpenAI and Google and, you know, Facebook, like, or Meta, I should say, whenever they talk about AI, they seem to show education use cases is like a major part of all their launches. I’m sure that’s not quite right, but it’s more than I can remember on most product launches outside of maybe the iPad over the last 20 years. And so it’s obviously getting a lot…education and AI together, Diane, are obviously getting a lot of attention and I find myself anyway, and we’ll talk about this in a moment. Like I start out with a strong prior and then I read a couple things and I completely flip my opinion and then I have that opinion and I talk to someone and then I change again. And so like I find myself pretty malleable still. But like you, it feels like this technology enabler that could be really, really intriguing. And we need to explore more on.

Diane Tavenner: I agree with you and I think we’ll do that in a way that we always do. We’re always looking for sort of third way solutions that are very practical and very pragmatic and very connected to what’s actually happening with young people in schools, with teachers. And so yeah, I think that, you know, people might be like thinking, oh my gosh, more AI. But I hope that we’re going to bring a, a sort of pragmatic approach to it that that is actually useful for people. 

The Teacher- v. Student-Centered Approach to AI

Michael Horn: Yeah, no, perfect. And I will tell you, when you visited my class and showed off Futre with the students, they noted that you never mentioned AI in your talk. So we are certainly not leading with AI, but we think it’s intriguing. And so against that let me start out with the opening framing I’d love to propose to you and then you can sort of react to how that framing sits. But it’s one that I’m stealing from a friend of ours in the venture world. And it’s something though that I’m noticing in the field and I don’t know that everyone who sort of is launching AI education products notices it this way. But what I’m seeing is that there’s sort of on the one hand a lot of AI startups and AI approaches that are very teacher centered or teacher facing as their entree, if you will, into the classroom or learning environments. And then on the other hand you have the student centered or student facing applications. This might be like the Khanmigo or, you know, some of those things that we’ve seen out there. And so there seems to me to be a bit of a dichotomy in terms of the startup space, the investors approach, different entrepreneurial approaches, even teacher, frankly and school designer and educator approaches on how they’re thinking about AI. Is it first a teacher tool or is it a student facing tool. What, what’s your take on that framing before we dig into each side of this?

Diane Tavenner: Yeah, so I think that sadly, and I will say sadly for me, I think most people are thinking about it from a teacher facing approach. And I think I sent you an article the other day or an op ed where I was like very frustrated with the premise, which was this exact premise. And I, as you know, I fundamentally disagree with that approach. Do I think we should be using AI as a tool to support teachers and to support students? Yes, but I think we’re just retreading the, the old way of thinking about schools. And let me just start, Michael, and say, like in this conversation, I’m almost exclusively going to be talking about high schools because I think elementary schools are quite different. And, and you know, so if we get into elementary school, let’s note that specifically. But for me, I’m very much thinking about high school, maybe middle school as well, but older students, and I just think that the world is going in a direction for many, many, many reasons where they need to be owning and driving their own education. Of course, this is not unique for me. I’ve been doing this for a couple decades at this point. This is my fundamental belief. There’s such a downside that we are not focused on how we enable students to own and drive their own learning. And AI is such a game changer. I think potentially in this direction it can help us do things we’ve wanted to do and can’t do. And we’re completely missing the mark when our total focus is on the teacher and how this is a tool that we’re going to build for teachers.

Michael Horn: No, that’s helpful. And right out of the gates. We know where you stand. I’m going to try to make the argument for the teacher facing up front and then you can throw cold water on me afterwards if you’d like. But, but let me try it. And, and maybe the way I will try it though is more to explain it about why I think the phenomenon is happening. And so number one, I would say on the why is AI better for teachers than students? Say false dichotomy, but let’s go with it. I think part of the approach is, look, AI hallucinates all the time. It makes mistakes. And these tools are better in the hands of experts rather than novices who can, you know, catch those mistakes and correct them in some ways. So number one, there’s sort of like a risk aversion approach to it. And so, and I think this, you know, we could probably contradict this in certain ways. But I think the AI like as risk to students is maybe driving some of this. Number one. Let me quickly add on that that I do think that there is something to it in the sense of AI when used by Amazon to get you to buy something that maybe you’ve looked at online. If they move the dial 0.001% that is serious dollars to their bottom line and if they alienate you, they don’t really care. Right. Whereas in education I think the argument would be if we actually mislead a student or you know, tell them a narrative about themselves that is going to mislead them in some, you know, like we could do deep damage to their self efficacy and, and, and, and sense of self and even their agency right down the line. And so that’s the reason for a teacher facing perspective maybe. Let me pause there before I go to the other two reasons because I, I that like a meaty set of claims that I think you should engage with first.

Diane Tavenner: Well, I think you’re uncovering one of the challenges that we have in education right now, which is a comp. Just a real lack of imagination about what is going to be possible because of AI. And so I think that many people, most, I don’t know, a lot of people at this point have logged on to you know, ChatGPT or one of the others and they’ve typed something in that little box maybe a few times and they’ve had or they’ve read articles about these hallucinations. But in many people’s minds like that is what quote AI is. Maybe some people now are playing with NotebookLM from Google. And you know, one of the really amazing things I think is that you know, it will produce a podcast.

Michael Horn: It’s pretty remarkable. A little over engineered but pretty remarkable.

Diane Tavenner: And it is like at first it’s like pretty mind blowing and then when you actually start to listen, yes, it’s getting all the right words. I did it the other day. Someone like loaded a chapter from my book into it and then it produced a 22 minute podcast. Man and a woman talking. And I was like. And they were like is this the conversation you and Michael would have had about your book? You know and like there’s pieces of it, yes, but it’s not us, it’s not human. It is, it’s like literally going, it’s read what is on the page and then it’s like making it, sort of bringing it to life. But there’s no thinking and nuance. And dynamism there. Anyway, all my point is that that’s not, you know, that product is the, is one of where they’ve taken what’s underneath it, the AI and they’re actually turning into something that is more user facing. So my assumption is that we’ve only just begun to see what’s possible. And so this idea that like, is that like chat box going to revolutionize learning for kids. No, it’s not. But that’s not what we’re talking about here. We’re talking about as a tool embedded in really well designed experiences in my view, products that will move the needle. And so I think you, you minimize or eliminate those risks that you’re talking about when you build it in thoughtfully. Certainly that’s what we’re doing on our team and so. 

The Sustaining Innovations of Teacher-Centered AI

Michael Horn: Well, and you’re leading with the product design as opposed to the AI, which is also a difference. Right. So let me say the second reason. I think that we’re seeing a lot of teacher facing things, which is that frankly, relative to today’s classroom, it does not require redesigning today’s classroom. It is in our language, a sustaining innovation relative to today’s classroom. And let’s be honest, that’s where the market is, right? As in, if you’re looking for volume, it is not in. I mean, yes, microschools are taking off, but there’s still a small percentage Of the education landscape. Certainly in the US, even more so in the world. And so teacher facing sort of as gateway in teacher directed instruction is where the market is. And frankly most VCs, when they enter a market, they have a five to seven year time frame to get out of the investment. They’re looking for unicorns within that. And that pushes you to where the dollars are, not where perhaps the puck should be going. So I think that’s the other thing driving this dichotomy, if you will.

Diane Tavenner: I think you’re right and I think this is the problem we consistently have every time we think something might help us transform schools, right. Is that the, it gets, it get the gravitational pull back into the box, the box of the school, the box of the classroom, the box of the teacher, the box of the course, it just, all the pullback to that is so strong and every time people try to unbundle it or disrupt it, we’ve had many conversations about that. You know, there’s a few outliers who sort of make it outside of that, you know, planet, orbit, gravitational. I spent a lot of time with a lot of them last week and it’s very exciting and inspiring with them and then you get back to the mass market is all still living inside that box. And so, so I mean, this is where I just feel like, I feel like I can’t help but get hopeful and excited, but I’m a little bit worried that I’m going to get my heart broke yet again about the potential changes that we might see. Because that’s what I want to have happen. I actually want to break apart that model and change this to be a learning experience, at least at the high school level, where kids are truly driving their own learning and learning in ways that are much more customized and personalized for them. And let me just be super clear, that does not mean they’re learning alone. This is still very group oriented. It’s actually quite real world oriented and that’s what I think is possible. So.

Michael Horn: But it’s not to say, let me just modify this before we jump to where you’re going, which is, I think you’d agree, there are plenty of low hanging fruit use cases to like to, to improve. Right. Teacher practice with AI, whether it’s better lesson plans, more diverse ways of reaching different student needs, et cetera, et cetera, et cetera, frankly, assessment, probably to get more real time information where your students are or how they’re doing to. Or simplify a teacher’s workflow. 

Diane Tavenner: Yeah, and that might be the middle ground here. You know, I was the other day, I was sitting there thinking through how can we disaggregate the role of the teacher and what does AI enable? This could still exist in the box model of class, you know, but I do think it would be an improvement. So if we think about all the hats a teacher wears, which are impossible. The job’s impossible as you know, I know everybody knows the feedback we’re getting from the market is it’s impossible because no one wants to do the job anymore. People will get mad that I said that. That’s not true. Some people want to do the job, but here’s the job. So one, you are. And these are the main things that teachers think about and people think about. You’re planning your curriculum and your lessons and you’re delivering them. There’s a real argument that AI, that a single individual teacher should never be planning their own curriculum again, ever, ever, ever, ever. It’s like not time well spent. It will never be as good as what can be done, you know, more globally and with all the learning science and expertise that we have, and even quite frankly, the delivery a lot of it is not personalized and individualized. So that could very much be, you know, AI driven, technology driven. Then there’s feedback and assessment.

Diane Tavenner: So I’m giving you like feedback. I know you’ve been grading some papers and assessing work. And again, like, again, we’ve done this for a decade plus at Summit where we took most of that off teachers’ plates. And the technology is absolutely capable of doing this now and better quite frankly, than humans. And so if we take that, that’s like the core of what most people think the teacher’s job is. So what’s left? And it’s the very human things, the things that I would argue matter. It’s the coaching and the mentoring of students. It’s helping them to figure out how they’re going to like sequence their learning pathway and what comes next and what happens when they get stuck and they need actual help in where they’re going. And so that coaching, that sequencing that, facilitating certainly a role in facilitating group learning and really cool real life learning experiences and giving real time feedback in those settings. There’s the social-emotional part of this. Like how do you, how do you become a person who understands a morning routine and actually, you know, knows how to manage your emotions and your relationships and all of those sorts of things. And then of course there’s like custodial care. That’s for younger, but to some extent older. Yeah, none of those things can be disrupted by AI, I do not believe. And for a lot of teachers, it’s the stuff that brings them real joy and it is really impactful for young people. So I think maybe the in between is a disaggregating of that role of the teacher.

Diane Tavenner: If I saw products moving in that direction, I’d be happy.

Michael Horn: So that that would be. So for all those listening, that’s the sustaining path we would like to see happen. And here’s the disruptive argument. Let’s get student facing here. Right. And student centered. And I think that is the argument. Right. 

Disruptive Applications for AI

Michael Horn: Is that yes, tutoring today or student facing tools. And I’ll get into the second use case in a second. But like the more narrow ones first. I’ve seen all sorts of critiques and I think we’ll get some of them on, on the podcast as we go through this series around how it’s not, you know, it’s yes, maybe procedural knowledge, but not like the in depth, really emotion driven. Right. Learning pieces and other things of that nature and it makes errors, you know. Right. All the rest. The Wall Street Journal has done a few hit jobs on things and so forth. But if you get into non-consumption where the alternative is nothing at all, I don’t have access to a tutor if I’m, you know, however many millions of kids in the United States, let alone the world, clearly better than the alternative, nothing at all. There are some very interesting places to launch student facing applications in that area, number one. And number two, I think the argument for it, and I think this is where you also might be going is I see it as lifting the quality of work of what students are doing because AI now is a tool of work just like we use it in our workplace to better…so that they can create more in depth, more exciting, you know, things. Right. With spending a little bit less time on some of the mechanics and more time on the depth, if you will, of learning and evidence right in the product or performance or whatever they’re creating. And I’m being somewhat vague because trying to capture all the possible use cases one could imagine depending on what subject or grade you’re imagining as we’re talking. But I think that’s the other area is that like the sense of agency for kids where they can actually build professional level skills stuff as they’re exploring.

Diane Tavenner: Yeah.

Michael Horn: Has just taken a big step up. And it’s not to say that they don’t have to learn the knowledge and application and skills. They do. But then using AI to level up all of that is pretty interesting, I think. Go ahead.

Diane Tavenner: Yeah. Let’s talk for a minute, Michael, about the broader context and why – because I think it’s so relevant here about what’s going on – I think in the world why this matters. So, so number one, it’s unequivocal. I just spent last week with people from the left and the right and like everywhere in between. And there’s a, there is an incredible agreement around the idea that school needs to be real world. It needs to be preparing young people, especially high school, for the real world, for jobs, for employment. It can’t be sort of this like theoretical, you know, thing anymore. And it’s not preparing them for that. It’s not preparing them with what I would just call basic professional skills. Like how do you actually like be an employee? How do you show up on time, how do you have agency, how do you do these things? And, and it’s, it’s not actually, if it’s not incorporating AI and how you use that in real work, it’s not going to be preparing them for the future that they’re walking into. And so I think that is happening. There’s a real move towards, you know, CTE, you know, career and technical education. As we know, we’ve got ESAs coming on in multiple states where people are going to be able to sort of more pick and choose their education. So you’ve got a lot of stuff happening where people are like, I don’t want to sit and get anymore and it’s not going to serve me to just sit there and take direction and then wait for you to tell me the next direction. And so I think what, you know. Do I think it’s a chat bot that’s tutoring me? You know, I think that’s super rudimentary. I think there’s so much better stuff coming, but you got to start somewhere. And I think that what’s more important to me is that like it’s, it’s breaking this dynamic of like 25 or 30 kids in a classroom, like waiting on instruction and the slowness of it and the exactitude of it and that. And so it’s moving us towards like this is the world we’re going to.

Michael Horn: The waiting on is a really particularly interesting place I’d love to like pick up on because I see the same thing, no surprise perhaps in that one we’ve been pretty clear that more connection to real world is important. I also think the ability to codify and create like standard curriculum, given the fast changing nature of real work is going to be a fool’s errand. And so that pushes you more and more in the direction you’ve been around. Experiential. Right. And so as a result of that, like that’s going to be doing, which means not like you can’t be waiting on the sort of the one scarce resource in the classroom to come over to you, unlock the lesson plan for you and then you’re allowed to go learn that. That that’s not going to be the model that engages or works, frankly. And so it’s everything from knowledge acquisition to exploration. On the one hand we’ll put that like as a big bucket, right? To actually engaging with, connecting with and then doing the work. And AI is a really interesting portal, I think, into all three of those, I guess is the way I would think about it. Whether it’s up leveling the quality of resources on the front end or frankly up leveling the level of work that young people are able to do and they’re showcasing of that and problem solving to real professionals and getting real feedback on it.

Diane Tavenner: Well, and I think this is so critical, Michael, because one of the things we’re seeing in the job market for, you know, post high school graduates, post college graduates is. And one of the things AI is doing is, is sort of competing with or removing those kind of entry levels. So no one wants to hire someone who doesn’t have experience anymore. You, you know, almost every job says you need a couple years of experience. So how are young people supposed to get experience? Well, their education is going to have to incorporate experience, if you will. It has to be experiential. It has to be a place where they’re going to be able to make the case that even though I just finished learning in some, you know, degree or credentialing program, I have experience. And so the, the act of learning and getting feedback and producing products has to be much more real world experiential if they’re going to have any hope of getting a job. 

Preparing Students for Success in the Workplace

Michael Horn: This isn’t an AI point, but I just, I’m, I love that we’re getting away from credential based hiring and that skills based hiring is a phrase, but I think I find it overly technocratic and a sense that we’re going to be able to define skills in narrow ways. And the word you just used, experience, to me is the way to think about it of experience based hiring. And the way you show you can do and step into a job is through the experiences you’ve had where you’ve done that. And if we believe, let’s go to the equity question. If we believe we want to give everyone a chance at that school has got to be providing it because otherwise my kids are going to be able to find those opportunities, but a lot aren’t. And so I think schools are going to need to be. A long time ago there was a professor, either I think UCLA, but maybe USC, and you can correct me if I’m wrong, and he wrote about how like schools of experience were the right way to hire people to see, like, you know, have you led teams, have you, etc., etc. As opposed to like, gee, Diane built a great product by herself, now we want her to be a manager. Two totally different sets of skills underlying that. Forget about naming the skills, let’s just look at the experiences themselves and say, like, how’d you do what, what lessons did you learn? What would you do next time? How does it equate to the culture here? Those are the sorts of questions and conversations I’d love us to be having in hiring. And so what you Just said, I think makes a lot of sense for the schools to be stepping into that. And the challenge, right, if we stay with our teacher centered model is that to ask teachers to sort of be the font of all of that is, is crazy.

Diane Tavenner: It’s not even, it’s not even possible by definition. You know, they are, they’re, they’re waiting on, the student is waiting on instruct. It’s not preparing them to be productive. No, no. And it’s not even neutral anymore. It’s negative because the incentive system in our traditional schools is actually counterproductive. It’s creating behaviors and incentivizing behaviors that are, are counterproductive when you’re going into the real world. And so, and I would argue the learning isn’t even that great. So it’s not like they’re coming out as masters of math or you know, and, and on top of it that. I mean, let’s just go back in time for a moment. We talk a lot about the industrial model and wanting to move away from industrial model schools. But I think some of the things that people forget is the design of the industrial model school was actually preparing people for…

Michael Horn: An industrial model economy.

Diane Tavenner: The factory. Like you showed up to a bell, you moved on a bell, you, you produced work at a rate and a speed in a way that was going to be very real world, very comparable to what you were going into. And our schools look nothing like workplaces at all anymore. And they’re not preparing young people for all of those pieces of it.

Michael Horn: No, we’re going off AI but I’m going to make one more point and then maybe we’ll bring it back. Which is, I actually think when people think about higher education also, and they’re like, oh, the rarefied university experience that I want all 55 million people for some reason to have, that is, you know, Harvard or whatever else, they forget that that is also a vocational experience, which is to train people for the professoriate down the line to prepare them to get master’s and PhD degrees. So for what it’s worth, I think it all has echoes to your point of the world into which you were trying to prepare individuals. And that world has changed totally.

Diane Tavenner: And I think that AI becomes a tool because I think a lot of the objections to changing this model, if you will, the box model, the classroom, the school building, etc, has been like, how can we actually do that? We have, you know, 55 million people in the schooling system. There’s a huge operational component like, how do you actually do that? And I do think that AI brings us a new set of tools in a very meaningful way if we deploy them properly. Not properly, if we deploy them, you know, interesting, smart, visionary ways that make that more and more possible.

Michael Horn: Maybe let’s leave the conversation there and I’ll put out one more question that I’m really interested to get from folks, which is when…We’re going to talk to people who are skeptics, who are optimists, probably in between. And the questions that I’m curious about are many. But one of them on what you just said is like, how does it maybe make certain things that we thought were important historically less so in the future? Like, yes, it might ruin the ability to do X, Y and Z, because AI is going to do it. But also that thing is no longer that important either as an artifact anymore. And where is that not true? Where is it going to ruin that thing that actually still really is important? How do we think about that? I’m curious to hear what people think.

Diane Tavenner: I’m curious about that too. I also will just put an invitation out, Michael. You know, we’re gonna do this for a little bit and we’ve got certainly a list of people we want to talk to and a list of questions. But we always love hearing from listeners. And so if there are people or questions you are curious about, send them our way and we’ll do the best we can.

Michael Horn: Perfect. Ok so let’s leave it there. Lots of, lots of energy around where we want to see AI solve problems. And let’s flip, as we always do, to what we’re reading, listening, watching, basically anything outside of our day jobs. What’s on your list, Diane?

Diane Tavenner: Well, I have one that’s legitimately outside of my day job, which is The Diplomat Season 2. And it’s just, that’s so bad.

Michael Horn: I need to get on that train. I really, for a variety of reasons, I know I would like it, so I will try to catch up to you. Mine is less, is, is not actually divorced from my work. I’m reading student papers non-stop right now. The AI I’ve tried a couple AI tools, Diane, that grade, I will tell you they don’t because they don’t understand context and the content knowledge. They’re very good at telling me, you know, grammatical things. I am not an English teacher. I don’t, I literally don’t care as long as I, it communicates the point in this particular case. So as a result, it’s still manual labor for me, for the next few days.

Diane Tavenner: Well, I’m so sorry. I hope that ends.

Michael Horn: No, all good. Some of them are great ideas, and I’ll hold to those. But for all of you listening, thanks as always. We look forward to hearing from you. Look forward to hearing your thoughts about who we ought to talk to, what we ought to learn from. We’re excited to do this and do a deep dive on AI with all of you. Thanks so much. And we’ll see you next time on Class Disrupted.

Get stories like these delivered straight to your inbox. Sign up for The 74 Newsletter

Republish This Article

We want our stories to be shared as widely as possible — for free.

Please view The 74's republishing terms.





On The 74 Today