upper waypoint

How Should Schools Adapt to AI?

We'll talk with educators and researchers about how AI is reshaping the classroom.
Balboa High School teacher Michael Rosenberg checks the text from an essay assignment to see how much was generated by AI. The text highlighted in yellow is identified as AI generated text. (Beth LaBerge/KQED)

Airdate: Monday, October 26, 2026 at 9 am

Artificial intelligence has long entered the classroom and is rapidly changing how students learn and teachers teach. Some schools are embracing AI and personalized learning tools, while others are pulling back on technology in the classroom. Meanwhile, students are increasingly turning to chatbots to help with assignments, raising concerns about what happens to critical thinking, reading, and writing skills when AI does the work for them. Can schools harness the benefits of AI without compromising students’ learning? We’ll talk with educators and researchers about how AI is reshaping the classroom, what it means for students’ development, and how schools and parents are navigating this new era of education.

Guests:

  • Daniel Herman, humanities teacher, Maybeck High School; wrote about teaching English in the age of ChatGPT for The Atlantic
  • Jeremy Roschelle, executive director of learning sciences research, Digital Promise
  • Kip Glazer, principal, Mountain View High School
  • MacKenzie Price, co-founder, Alpha Schools
  • Tom Wadbrook, director of technology, Prospect Sierra School

Episode Transcript

This is a computer-generated transcript. While our team has reviewed it, there may be errors.

Alexis Madrigal: Welcome to Forum. I’m Alexis Madrigal. Current-day generative AI models can write just about everything, and we really haven’t wrapped our heads around having machines that can write. Some of us are ignoring this new reality or finding ways to minimize the power of these models. Others are enthusiastic or overenthusiastic adopters, happy to be done with one of life’s chores or delighted not to have to write in a second or third language.

For professional writers, there’s a lot of well-founded fear and anger, and maybe in some people, a dark tingle of excitement. But what about the educators who are supposed to teach our children? Children who, after all, are often tested via writing assignments that any model could ace in 10 seconds. How is the task of teaching different in this new world? And how should school administrators respond?

We’re joined first by two educators, a teacher and a principal. Daniel Herman is a humanities teacher at Maybeck High School in Berkeley and a faculty associate at Bard College’s Institute for Writing and Thinking, who has written a couple of great essays about teaching English in the age of ChatGPT for The Atlantic. Welcome to Forum, Daniel. Good morning.

And we’ve got Kip Glazer, principal at Mountain View High School, author of Ready to Lead with AI: A Practical Guide for School Leaders. Welcome. Good morning.

So, Daniel, I went back to the essay you wrote shortly after the launch of ChatGPT. You wrote this essay, “The End of High School English,” and it raised a lot of questions about the future of teaching writing in high school classrooms. So, four years later, how has AI actually changed things in your classroom, and what are you seeing from students?

Daniel Herman: In many ways, very little has changed. My classroom remains pretty much the way that it’s always been. I remain committed to the idea that the classroom is a sacred space. bell hooks, I think, referred to it as a location of possibility. It’s a place where people come together to figure out a problem, explore an idea, hear different perspectives, and make meaning together, and where there’s a teacher who can model the behaviors, intellectual and otherwise, that they want their students to emulate.

It’s when we turn to assessments that things can get a little dicey. So I’ve definitely had to go back and recalibrate or reassess all the things that I ask my students to do, not to make them AI-proof in the sense of preventing them from cheating, but really making sure that the things I’m asking them to do are in line with this new reality, where there is this computer that can do so many of the things that we used to assume students had to know how to do.

So I’ve definitely encouraged my students’ writing to be way weirder than I ever did before, but overall —

Alexis Madrigal: My emails have gotten that way, too. Sorry for everyone out there. You’ve got to prove you really did it, you know?

Daniel Herman: Overall, I’d say that it’s a question of emphasizing process over product. So instead of writing a research paper, maybe we’re doing research presentations where we’re sharing our learning together, or an interview project, but showing that we’re thinking on the page and really grappling with a problem.

Alexis Madrigal: It also really feels like, for you, the introduction of these models has been a catalyst for thinking about the differences between what it is to think, what it is to learn, and what it is to assess that someone has learned. That seems to be how you’ve taken it as a prompt.

Daniel Herman: Totally. My rule of thumb, and this is maybe a little bit controversial, is that AI is actually pretty useful for learning, but terrible for thinking. I think the thing that took me a while to really figure out is that, for a 16-year-old, thinking for themselves does not go without saying. It really, in many ways, can come to the teacher to inspire that inquisitive impulse and persuade them that that sort of thing teachers sometimes talk about as productive struggle is actually worth their time, rather than just getting the lowest-common-denominator response to a problem.

Alexis Madrigal: So, Kip, you’ve got a different set of challenges. You’re principal of Mountain View High School, in the heart of Silicon Valley. What does this look like at your school, with your educators and your students?

Kip Glazer: That’s a great question. I think, as a school leader, we want the student outcome to be what we desire it to be. And the environment sometimes doesn’t allow educators or school leaders to do what we know to be correct, because there’s so much noise around what is the right way to teach or what is the right outcome as a result of learning.

Our students are grappling with that. Some students actually say, “I don’t want to be defined as a cheater, so I’m going to never use this tool.” And that all-or-nothing thinking is, I think, really troubling for me.

Dan and I were talking about it a little bit. We need to really think about how, as adults, to sit in the middle of the messiness, as much as we want our students to do that. And how do we model that? It’s really a challenge when we ourselves don’t really know how to interact with this new reality.

Alexis Madrigal: That’s an interesting point. Daniel was talking about modeling the intellectual and other behaviors he wants to see in his students. So what is the right thing for teachers, then, to model in their use of AI? Or for administrators to model?

Kip Glazer: Yeah, I think about the identity threat, and I think that educators —

Alexis Madrigal: What do you mean by that? Spin that out a little bit for listeners.

Kip Glazer: Yeah. For the longest time, I think educators have felt this sense of responsibility to be the most educated and learned person who can guide the way for our students, who are trusting us to follow and go, “Where is that intellectual pursuit?”

And when there is a tool that can actually mimic that behavior, sometimes more sophisticatedly or more refined than we can, there has to be a way of feeling threatened in terms of how we’re supposed to show up.

I think some teachers’ way of dealing with it is to ban it all and say that we’re protecting ourselves in our insecurities or our struggles. Other teachers I see are able to share that they don’t know it all, and it’s OK for us to discover it together.

And I see responses to that sort of honesty and vulnerability in two ways. Some students are actually embracing that, but other students are like, “When you don’t feel secure in what you know, how am I supposed to feel secure in how I’m supposed to follow?” Yeah, I’m supposed to follow. And I think there is a lot more work to be done in terms of how we emotionally support this transformation in our classrooms.

Alexis Madrigal: You know, Daniel, I have an East Bay 13-year-old who despises AI and is just like, “We’re destroying the world for this stupid technology.” What do you find with your East Bay teenage students about their relationship to this technology?

Daniel Herman: Yeah, that’s funny. I mean, I don’t feel like there’s much curiosity about the technology. You know, this is just the world that they’ve inherited. They’re the ones who coined “AI slop.” This stuff has flooded their TikToks. This is their reality. This is the world they’re swimming in.

I totally agree with everything that Kip says. We really have to make sure that this is a dialogue, and I really appreciate you mentioning vulnerability. I ask my students for honesty and transparency. I think we really need to take seriously that this technology is arriving in a world where our students, like all of us, are stretched thin in so many ways: busy and overwhelmed, mental health crises, the whole thing.

And now there’s this magical tool that can do their homework for them, and we can’t blame them if some of them are tempted to use it. So I really ask my students to be open with me if they’re thinking about using it, because depending on the class, there are definitely ways that it can be useful to their thinking and their learning, and other ways that it entirely gets in the way of their thinking and learning.

Whether all my students have abided by that request, I can’t say, but it definitely has to be a conversation.

Alexis Madrigal: Kip, what do you say to students who you find out are, in fact, using chatbots to do their homework? Or even, at this point, we’re not just talking about chatbots. They could be using agents to — I mean, you were talking before the show about a student who wrote an agent to go download the homework from Canva, do the homework in their own hand, their own handwriting, and then upload the assignment.

At one level, you’re impressed, but in that kind of Ferris Bueller way, where you’re like, “Oh, God, I can’t believe you did that.” Although, kind of impressive, but gah, I can’t believe you did it. So what do you do with a student like that?

Kip Glazer: Yeah, and that’s why I think what we’re defining as learning and how we’re assessing it has to be at the forefront of every conversation.

I think the AI slop, or accelerating assessment, happens because the current platform invites that, right? When you have these worksheets that are being mass-produced, students know that. And one of the things that I hear from my students is, “How come teachers can replicate or not even grade my work and use AI tools, but I’m not supposed to use it?”

And when you’re working with teenagers, it’s beautiful to hear. They can spot hypocrisy a mile away, and some of them, politely or not so politely, call that out. I think we have to be honest in terms of how we’re allowing our work to be accelerated and what that means for our students.

One of the dangers of these tools is that we’re not invited or allowed to be present anymore. We’re always trying to accelerate. Faster and more is always better — to what end?

I’m getting emails from seventh graders wanting to know what they have to do now so that they can be successful in high school. Two years ahead, they’re really trying to plan ahead. On the one hand, I’m impressed, but on the other hand, this is the situation we’re creating. Acceleration to what end?

I think that is actually also really dangerous because when you deeply learn something, you need to take the time and be patient with yourself, as much as the world, and we’re not allowing our students to have that moment.

Alexis Madrigal: I mean, it just seems like a lot of kids take the task to be good at school, not to sort of learn, right? And I understand that. I remember being a kid and thinking that, too.

We’re talking about artificial intelligence in the classroom. We’re joined by Daniel Herman, humanities teacher at Maybeck High School. We also have Kip Glazer, principal at Mountain View High School, and, of course, we want to invite you into the conversation.

To parents: How is AI showing up in your kid’s classroom, or maybe just on your kid’s phone? What are your questions or concerns about this era of learning?

And, you know, we know maybe there’s some teachers out there with a prep period. How is AI changing the way you teach, or how your students are approaching their learning?

Give us a call at 866-733-6786. That’s 866-733-6786. Or email us at [email protected].

I’m Alexis Madrigal. Stay tuned.

Welcome back to Forum. Alexis Madrigal here. We’re talking about AI in the classroom, trying to move this conversation forward a little bit. I’m joined by Maybeck High School’s Daniel Herman and the principal at Mountain View High School, Kip Glazer.

We’re going to add a few other guests into the conversation here. We’ve got Jeremy Roschelle, executive director of learning sciences research at Digital Promise and co-host of the Future Fluent podcast. Welcome, Jeremy.

Jeremy Roschelle: Great to be here.

Alexis Madrigal: We’ve got MacKenzie Price, co-founder of Alpha Schools and Two Hour Learning. Welcome.

MacKenzie Price: Thank you for having me.

Alexis Madrigal: Yeah, thanks for joining us. And we’ve got Tom Wadbrook, who’s director of technology at Prospect Sierra, a school in El Cerrito. Welcome.

Tom Wadbrook: Thank you very much.

Alexis Madrigal: So, Jeremy, your perspective as a learning scientist: What needs to happen for students to actually learn something? What do we know about that? And do you think that AI actually disrupts that ability for students to learn?

Jeremy Roschelle: Yeah, great place to start. And I just love the focus on learning and not just on this technology. The antidote to too much technology is to talk more about learning. What do we want learning to be?

Daniel painted a great portrait of his classroom. The students are active. They’ve got something constructive that they’re working on together. They’re building knowledge as a group, and they’re social. Those are three absolutely critical ingredients to any successful learning environment.

Alexis Madrigal: Why is it, then, that I’m hearing that if students use AI, their brain function falls off? This seems to be a really serious concern of many educators and parents.

Jeremy Roschelle: Yeah, let’s go back to those three words. Look at what the students are doing. Are they active, for example, in solving the math problem? Or is it copy and paste, and it’s doing it for them? If their brain is not active, they’re not learning.

Are they misdirecting time away from conversations with their peers and teachers to just be talking to a technology? Well, social interaction is falling by the wayside.

And Kip brought up, you know, is it just trying to race to get done? Whereas Daniel brought up process, and that’s the constructive part. Are we paying enough attention to the learning process and kids actually making things? That’s how you learn. Or are they just racing to get an output?

Alexis Madrigal: So interesting. MacKenzie, at Alpha Schools, your model centers on two hours of academics a day using AI-powered learning tools. Do you see that as sort of a race to get done? Or how do you interpret what students are doing during those two hours? How do you guide them?

MacKenzie Price: I will tell you, I think there has never been a more exciting time to be a five-year-old. And I think what we’re seeing in education is seriously an incredible moment where we can finally allow students to get the individual, personalized attention that is impossible for even the best teacher to provide to 15, 20, or 25 students in a room.

And that’s what we’re doing at Alpha. We’re using artificial intelligence to create personalized learning plans that are able to meet the student where they are, at the level they need to be.

As a result, during the two hours in the morning that Alpha provides academic teaching, our students are able to learn twice as much as students who are sitting in a traditional school classroom for the majority of the day.

And it really is because of two things. Number one, receiving that one-to-one, personalized, mastery-based tutoring experience. But number two, and most important and critical, is that the adults in our classrooms, the teachers, are freed up to spend time on what only humans can do, which is connect with the students, mentor them, coach them, and provide that motivational support.

That’s why we’re seeing such incredible results, because we’ve finally freed up the adult in the classroom to do what they’re great at.

Alexis Madrigal: MacKenzie, how do you evaluate that they’ve learned, quote, “twice as much” as other students?

MacKenzie Price: You know, it’s an amazing claim, and as a result, people are always wondering, “Well, how do we actually know this?”

What we use is NWEA math assessments, where we’re basically able to see an anticipated score that a student is expected to grow by in any given year. About 11 million students across the country take the math assessments, students from public, private, and home schools.

What we expect at Alpha is that our students’ RIT scores will go up twice as much as expected. So a student, let’s say, who’s in the 60th percentile in math in grade five, would be expected to go up three points over a given year. At Alpha, our commitment is that they’ll go up six points, and that’s what we’re seeing.

Again, this is such a new time in education, and it’s so different from what we’ve known for the last 150 years: a teacher in front of a classroom, kind of an everyone-teaching-at-the-same-pace model.

It’s important that we be able to show these results. I think it’s part of what our parents appreciate: seeing that students are learning so much better, retaining information, and then, of course, having so much more time freed up in the day to focus on human development and life skills.

Alexis Madrigal: Tom Wadbrook, let’s talk about the direction that Prospect Sierra in El Cerrito is taking. It serves students from TK through eighth grade, so there’s a very different culture around the use of AI. What are you hearing from parents and teachers at the school as you try to figure out how to approach the role of AI in a TK-through-eighth-grade classroom?

Tom Wadbrook: Yeah. I mean, I guess the first thing I would say is we have the luxury of being TK through eighth grade, not high school. We don’t have the same pressures, and so we can take it slower. But we really are taking it slowly.

We’re very intentional as a school, very process-driven, and also collectively have quite a lot of, I wouldn’t say antipathy, but —

Alexis Madrigal: You’ve heard some things from parents.

Tom Wadbrook: Yeah, and teachers. I mean, to be honest, two or three years ago, when we first started talking about AI as a faculty, it was a frosty reception. I think part of that was fear of the unknown, and part of it is what you were alluding to earlier, Kip, which is, you know, what is my identity?

Maybe we’ve moved on from fear and antipathy and indifference to some curiosity and exploration of AI in the classroom from our teachers, but there’s still a lot of indifference and quite a lot of antipathy.

Alexis Madrigal: There’s some key context here, too, that I’ve heard you talk about, which is that during the pandemic, of course, everything went super digital. All the assignments were online. Everyone was on a Chromebook or an iPad or whatever.

How is Prospect Sierra thinking about the overall amount of technology that’s present in education, with AI sort of wedged in there or coming in like a meteor strike into that context?

Tom Wadbrook: Yeah, I mean, that’s great. If you put AI aside for a second, the pandemic happened, we put devices in every kid’s hands, we sent them home. It was a mixed bag at best, right? Everyone came back, and we kind of kept on doing the same thing for a year or two. Then we realized that that was not a good direction.

So we’ve been on what we’re calling a tech rollback. We’re in year three now, maybe even year four. And it’s really just been incremental. We don’t let the iPads and laptops go home in fifth through seventh grade anymore. They stay in the classroom. They get used a lot more intentionally. We’ve dialed back all of the apps that might be on them.

So, yeah, we’re really trying to bring back down the level of technology that we’re using.

Alexis Madrigal: There’s a parent faction that’s saying, “Well, the kids already have access to all these tools. As adults, they’re going to have access to all these tools. We must teach them to double down, AI-max, because that’s the world they’re going to live in.”

Tom Wadbrook: We do. I mean, we have the full gamut, from “There shouldn’t be any AI” to “You should be doing absolutely everything you can in terms of teaching these kids to immerse themselves in AI.”

So we’re figuring out where we fit. As I said, we’re being very slow and very intentional about it. We’ve basically figured out that there are a few baseline things we need to do. We need to teach them about wellness and safety, academic integrity, so cheating, obviously, and to ignite — either when they’re using AI or when they’re consuming AI.

We’ve kind of set that as our baseline. And so now the big question for us is how much more AI belongs in the classroom. Is it none? Do we follow the model of New York schools? Or is it all the way in?

It’s not going to be all the way in. It’s going to be somewhere nearer the New York model, I suspect, but that’s really the big question we’re facing today.

Alexis Madrigal: Kip, do you have some AI-maxing parents who want you to double down on it, or do you actually find that, in the heart of Silicon Valley, it’s kind of the opposite?

Kip Glazer: I’m going to say that we’re in the same situation as Tom is in. I think the more we think about student wellness and the pressure on our students, our parents are actually pretty reasonable in terms of what they think.

And I think that’s the goal, right? Discernment. We talk a lot about that. If you’re going to use this tool correctly, you do need to go slow to go fast.

The more we can involve our parent community in the conversation, because ultimately our goal is to support our students, whether the school is doing it, a teacher is doing it, or parents are doing it, the more grown-ups are having reasonable conversations, the better our students will be in the long run.

Alexis Madrigal: Let’s start bringing in some calls. We’ve had four phones going since the second we opened them. Allison in Berkeley, welcome.

Allison: Hi, thank you so much. Yes, I’m part of a parent group in Berkeley called Berkeley Unplugged, and we’re really interested in this and related topics. A lot of great points have been made, and I just wanted to sort of emphasize two of them.

One is that, you know, I think when we’re talking about AI in schools, sometimes we lose sight of how AI is very different for a kid who is learning and whose brain is forming, who does not know how to do, for example, critical thinking, doesn’t know how to write yet. That’s very different from if you or I, as grown-ups, want to use AI.

So I think when we’re thinking about using AI, we should really think about it mindfully in school, so that we’re very cautious about kids so-called never-skilling and never learning how to do these things. I think that’s really, really dangerous.

The other point I just wanted to bring up is that teaching how to use AI appropriately is very different from allowing kids to just use AI for assignments and things like that and sort of skipping the whole process.

Essentially, in some ways, it raises the opportunity for them, as was said, to skip the learning process. Big picture, what are we doing here? Kids are there to learn through iterative failure, essentially, and through challenge. AI allows them to bypass that, and I think that’s really dangerous. We need to keep that in mind.

Alexis Madrigal: Allison, great points. Let’s go to Jeremy Roschelle, our executive director of learning sciences research at Digital Promise. What do you think about this concept of never-skilling and how to think about appropriate AI use or literacy at different ages?

Jeremy Roschelle: Yeah, thanks for the question. And I love everything that Allison brought up about the developing brain, because that’s what learning science is all about.

I think we have to be careful that AI is not all one thing. So let me give you two quick scenarios.

One: Allison was talking about kids’ ability to monitor what they’re doing as they’re using, for example, a chatbot to help them answer a question. If their brain isn’t fully developed, they may not be able to pull back and monitor their own learning and make sure they’re using it effectively.

Scenario number two: A child is learning to read and needs a lot of decoding support. An AI-based technology can listen to them read a real book, and when they miss a word, help them with the pronunciation of the syllables they missed.

I don’t feel the developmental risk with the reading example, because it’s not an open-ended chatbot that’s answering questions. It’s skill training for how to decode words.

So AI is many different things. It can be used in different ways, and we have to analyze the use case, not the technology.

Alexis Madrigal: Yeah, yeah, that’s interesting. MacKenzie Price, what do you think about our Berkeley listener Allison worrying about skipping the learning process?

MacKenzie Price: I think it is a very valid concern. We have to make sure that AI does not enable children to outsource thinking. In fact, making sure kids get knowledge into their brains is more important than ever.

Teaching students how to build the skill of learning how to learn is absolutely critical. Not only do we need to make sure that our academic rigor is as high as ever, but we now have the opportunity to make it even better because of personalized education.

We also have to teach our students how to use these AI tools in ways that allow them to be creators and contributors, not consumers and not outsourcers.

Alexis Madrigal: What access to tools do they have in your classroom? Do they just have your platform, or can they have Claude running on the side?

MacKenzie Price: No, they definitely can’t. In fact, in the morning, what we often tell our students is, if they’re using AI in the morning, they’re probably cheating. And if they’re not using AI in the afternoon, they’re probably missing out on some sort of learning opportunity.

During that two-hour core learning block, the AI is really operating behind the scenes, simply assessing and understanding what lesson to put in front of a student. Our schools do not use chatbots. Chatbots in school are cheap bots, and they are not something that you should have in school.

I do like the direction that so many schools around the country are taking, which is, if they don’t know how to use AI well, then don’t give kids unfettered access to it.

However, what we see in the afternoons, as we’re helping students build the skills to be adaptable, become experts, and learn how to learn, is that it’s great to give them supervised access to some of these incredible tools.

We see middle school and high school students building businesses using things like Claude, turning profits and making significant amounts of money on a regular basis. We also see kids taking their passions for art and music and using what they’ve built naturally to expand and grow using technology.

I think that’s a skill that’s going to be so important for our young people to have when they go out into the world, because we’re no longer in a situation where everything we need to know we’ve learned by the time we graduate college.

So we give our students a lot of opportunities to get to know how to use these tools well and go from there. But, you know, I enjoy thoughtful dialogue like what we’re having, because you do have to make sure it’s done right. Having technology in a classroom for technology’s sake is not the right thing to do.

I think it’s part of why we’ve seen a major degradation in the quality of our education. So I appreciate what this mom and the group have said around doing that, but we’ve got to make sure that screen time is done intelligently. We have an incredible opportunity with AI when it’s done the right way.

Alexis Madrigal: I’m going to just give a couple of representative parent comments here coming in.

Mark writes, “I have four kids under seven. Let’s teach our kids how to use these tools! Too much Luddite fearmongering out there. The future is bright for our kids.”

Alex writes on Discord, “I really don’t think schools need to adapt to AI because using AI doesn’t require any skills that are new. Being a good reader, communicator, and critical thinker are what makes you good with AI. The best way to learn these things is the old-school ways of being forced to think for yourself and struggle. Paper problem sets, tons of book reading, written or oral exams. AI is what you use when you don’t care to deeply learn much from your process, and the neurological research reflects this.”

Obviously, a lot of strong feelings about this particular topic out there. We’d love to hear yours. How is AI showing up in your kid’s classroom? What are your questions or concerns about this new era of learning?

Teachers, if you’re out there, we’d love to hear your thoughts as well on how AI is changing how you teach or how your students learn. You can give us a call. The number is 866-733-6786. That’s 866-733-6786. The email is [email protected]. And you can find us on social media: Bluesky, Instagram, or Discord, at KQED Forum.

We are, of course, talking about AI in the classroom, and we are joined by a big panel here. MacKenzie Price, co-founder of Alpha Schools and Two Hour Learning. We’ve got Tom Wadbrook, director of technology at Prospect Sierra up in El Cerrito. We’ve got Jeremy Roschelle, executive director of learning sciences research at Digital Promise. And, of course, we have two educators: Kip Glazer, principal at Mountain View High School and author of Ready to Lead with AI: A Practical Guide for School Leaders, and Daniel Herman, a humanities teacher at Maybeck High School and faculty associate at Bard College’s Institute for Writing and Thinking.

I’m Alexis Madrigal. Stay tuned for more right after the break.

Welcome back to Forum. Alexis Madrigal here with a great panel. We’ve got Maybeck High School’s Daniel Herman, Mountain View High School’s Kip Glazer, executive director of learning sciences research at Digital Promise Jeremy Roschelle, MacKenzie Price, co-founder of Alpha Schools, and Tom Wadbrook, director of technology at Prospect Sierra.

We are, of course, talking about AI in schools and all of the challenges and opportunities that it’s presenting. Let’s bring in some more callers. So let’s go to Megan T. in San Francisco. Welcome.

Megan T.: Hi. Good morning, everyone. Thanks for this conversation. I work at the Exploratorium, and we’ve worked to empower teachers to transform classrooms for a long time.

I love the question “What are we doing here?” That’s one of the right questions. I agree with some of the guests that there’s an exciting opportunity here. But I don’t think it’s about personalization or even whether or not we use AI.

I think if we’re going to keep bringing humans together in classroom spaces every day, we better be sure what we’re doing with those humans and bringing them together for reasons that matter.

I’m curious what folks think the purpose of the classroom should be next in an era of AI, but also climate change and societal change. I’ll take my answers off the air.

Alexis Madrigal: Megan T., thanks so much. And we’ve got to go to our teacher here, Daniel Herman.

Daniel Herman: What’s it all about? I don’t know. I mean, I guess I’ll say I’m really skeptical about learning for learning’s sake. Teaching somebody the orthodox view of an Emily Dickinson poem — I don’t really believe there’s any inherent value there. The point is to have a meaningful connection to the subject, whatever it is.

Writing is really sort of the crucible where this happens. Five years ago, I would have told you that writing is an absolutely essential skill, uncompromisable, that everybody needs to know how to write. Now I would say that writing is a tool. It’s a technology itself, a powerful, wonderful, liberative technology, but really just a way to grapple with complex ideas and organize our thinking.

In most of my classes, we spend most of our time writing together, writing in spiral notebooks, me included, and then sharing our writing around the table. They can take that writing home and use it to write something bigger.

At the same time, can I tell my students that in this age of TikTok and Instagram Reels, crafting an immaculate paragraph is as beneficial as being able to do video editing? Especially, I think about Zoran Mamdani’s short-form content. Can I tell them that being able to write, rather than make a beautiful video, is the best way to affect meaningful change in the world? I’m not sure that I can tell them either.

Tom Wadbrook: Can I add to that?

Alexis Madrigal: Yeah, go ahead, Tom.

Tom Wadbrook: I would just say that, specifically in my school, we’re very focused on human-to-human connection. So if you put aside learning for a moment and ask what a classroom looks like in the age of AI, what’s really important to us is teaching our kids self-awareness, empathy, and connection.

AI has got nothing to do with that. It’s all about adults and children being in a room together, connecting, and learning to be good humans.

Kip Glazer: Yeah, I think when we define learning as acquiring knowledge into your brain, we’re missing the foundational understanding of learning. Learning has always been social. It will remain social.

So what is a classroom supposed to be for? That’s the question I ask myself, as well as my students. What problem are we trying to solve, and for whom are we doing this? Just because we could doesn’t mean that we should. Knowing the difference really is human, and we have to lean into it.

This conversation about how we effectively improve learning is important, because no one wants to sit around and feel like they’re wasting time. But that brain rest, and what we think of as play, is where our students and children can really thrive.

I would ask all the adults who are trying to push our kids to be the version that we wish we had: we were different. We were slower. We were more playful. And I think our students are screaming for it. That’s why we have such a crisis with mental health and emotional dysregulation.

If we want our students to be healthier, happier adults, then we also need to have some self-reflection about what we’re pushing toward.

Alexis Madrigal: I really think that point about knowledge being social, too, is such a — I mean, obviously, in this job, you learn a ton, but you learn it in this very particular way of talking with all these people. Over time, the web of knowledge that you have is embedded with all these people across the whole community.

It’s a lot easier to remember stuff when you know who said it, why they said it, and the context of that person, rather than, you know, I read one sentence in a backgrounder that Claude created.

Let’s bring in Patricia in Redwood City. Welcome, Patricia.

Patricia: Hello. So, yeah, I’m a retired teacher, and I found in the last year or two I was teaching that more and more students were using AI to write their bibliographies for them. But they couldn’t tell the difference between whether it was correct or not, because they had never really taken the time to learn how to do it themselves.

So I felt like one way of dealing with AI in the classroom is simply to highlight its deficiencies, or its limitations.

We had contests where we would divide up the class, half using AI and half using physical books, and there was a time limit. I flashed the name of a book on the screen, and the first team to feel confident that they had a correct bibliographic entry would put it on the board.

And guess which team always won? It was the team using the books, because they actually took the time to get it correct from the book that was being used, whereas the AI team always had something that was not accurate.

Alexis Madrigal: When were you doing this, Patricia?

Patricia: This was only just two or three years ago.

Alexis Madrigal: God, I would love to — I want you to rerun it. I want to rerun it. Patricia, what an exercise. What a fun exercise and an interesting exercise.

The thing that it makes me think of, actually, is when AI companies are testing these tools, they do this thing called red-teaming. They’ll literally do exactly what you’re talking about, but they’ll tell the teams, “Try and go do something bad.” What they’re looking for is uplift.

If you want to build some bomb thing, you could find sources in libraries and on the internet that could help you do that, or you could use AI to try to do that as well. They try to look for the delta between the two teams. If a model is too good at helping them do something bad, then they try to —

So you kind of came up with the red-team exercise for your students, which I love.

One thing I wanted to come to on this, Daniel, is: What do we do with technology that’s evolving really quickly? Because I love that this happened for Patricia and her students, and they were able to identify some of these problems. But if we look at how things are going, maybe now, maybe in 2030, the students using AI will probably start to win. Then what do you do?

Daniel Herman: Yeah, it’s really been wild over the past four years to travel around and talk to teachers around the country. Teachers are a defensive group, and we do things a particular way, and we have every reason to want to continue doing things that way.

I’ve definitely noticed, over the years, teachers again and again trying to plant their flag in something that they want to be stable, and then the sand just slips away underneath them.

Immediately after ChatGPT was released, I saw a lot of teachers saying, “Well, it can’t even write well.” And then a couple of years later, “Now it can write well.” They say, “Well, it can’t even do research,” or “It can’t even integrate quotes,” or “It can’t do MLA citations.”

We have every reason to believe that every single bar we establish that it can’t jump over is going to be something we have to rethink. It’s about being flexible in our thinking and being willing to let go of something.

Now, I’m sure I’m going to make a lot of enemies out there, but I’m going to pick one stupid example, and that’s apostrophes.

Every English teacher will know you’ll get a student in September, and they struggle using apostrophes properly. You show them in September, and you show them in December, and you show them in February, and you show them in May, and it’s hard, you know? I mean, it’s really complicated. When it’s plural, it’s different. And apostrophes are just this made-up thing. It’s this little squiggle mark.

Then, to go back to Emily Dickinson, they read an Emily Dickinson poem, and she uses them in all kinds of weird ways.

So, for me, if a student wants to take their essay and put it into Claude and say, “Make sure my apostrophes are perfect,” and they never have to think about it again, and I never have to think about it again, fine.

I’m way more concerned with what Kip is talking about, and about empathy, and listening to people who we disagree with, and people bringing their ideas, experiences, and beliefs, and finding a way to be together with other human beings. That’s way more important than apostrophes or, forgive me, MLA citations.

Alexis Madrigal: A really interesting comment here. Nina writes, “I remember when people were debating whether children should be able to use calculators in school. The calculators won.”

So you think you know where this comment is going. Plot twist: “And now we have generations of people who can’t do simple arithmetic in their heads. For example, have you ever tried to cashier why the bill is wrong? They don’t have the basic math skills.”

The question there for me is, was it wrong to introduce the calculator into the classroom or not? I think most people would argue it’s OK to use calculators, given that adults use calculators.

But what do you think, Jeremy? When a new technology comes into play that can obviate the need to learn some skill, how do we say, “You must learn it anyway,” or “You shouldn’t have to learn it”? What’s your take on that?

Jeremy Roschelle: Yeah, graphing calculators are something we actually have a tremendous amount of evidence on, so I feel really happy right now, because it’s research that speaks to this.

Look, a great thing about that technology was that a good teacher could say, “Kids, calculators down. We’re learning basic math skills now.”

But take another scenario. Suppose you’re trying to learn what the slope of a graph means, and your alternative is pulling out graph paper, tediously plotting point by point, getting out a ruler, and drawing a line through it. By the time you go through all the steps, you’ve forgotten what you’re trying to do.

Whereas with a graphing calculator, you can see that slope shift as you change the equation you’re working with. That’s invaluable to kids’ understanding of the meaning of slope.

So I just think about how we design and implement the learning activities. And, again, we need educators, we need teachers who are prepared to hold the line. What are we trying to learn right now? Is it a basic skill? Put away the technology. Is it something where cognitive offloading is actually useful because it’s lower-level and we want to focus at a higher level right now? Go ahead, use the technology.

Alexis Madrigal: We have some comments coming from parents that I think get at the real dilemmas of this.

Leah writes, “Managing technology is the single worst part of modern parenting. AI is not hard to learn. Why do we need to teach it to our students? Learning how to learn is the point of education. Why do we want to make everything easy? If our kids learn that answers are at their fingertips at all times, they also won’t know how to struggle through non-technology parts of life, like learning how to knit, or work out, or master a sport. Isn’t that the point of a growth mindset?”

Daniel in San Mateo, somewhat relatedly, says, “What about the simple fact that since the proliferation of screens and tech in classrooms, growing evidence that AI degrades critical thinking, student attention disorders are up, and test scores are down? Why shouldn’t those results speak for themselves?”

Tom, are these some of the things you’ve been trying to think about as someone who’s considering the overall package of technology that kids are presented with? How do you help parents with “tech is the single worst part of modern parenting,” which I don’t disagree with?

Tom Wadbrook: Yeah, what do you do? I think a few things. My first response is that the really simplistic answer to both of those isn’t — it’s not about if you use AI; it’s about how you use it as a student, right?

I think, Jeremy, you could probably speak to this better than me, and maybe I’m wrong, but it seems to me that the evidence is coming out that if you teach a kid how to explore the subject they’re trying to learn using an AI tool, they can actually learn more than they might have without the tool. Versus if you give a kid AI and just let them find the answers, they’re not learning anything.

So, not if, but how.

Alexis Madrigal: Possibly. Mm-hmm, mm-hmm. Yeah, Katie in Martinez writes, “Will AI need to follow the path of media literacy, although I’m not sure media literacy worked out well? But how to interpret it, how to use it, how to understand critical thinking skills.”

Is that kind of where you’re headed at your school? At a high level, you think?

Kip Glazer: Yeah, we have an AI scale from zero to four, where zero means no AI and four means complete automation, with everything in between. We’re trying to be a little more intentional and focused on how we want to see the behaviors of the students.

I think good, sound pedagogy, being able to know when to jump in and when to back off, is really important.

Teachers today may not have the opportunity or even access, because we’re conflating exposure with expertise and exposure alone. The conversation around AI is that if you’ve used it once, then you already know how to use it as an expert.

But we all know as educators that experimentation and experience have to be in the middle to really get to the expertise level. We have to be patient with ourselves as well as our students, to be able to provide that opportunity. I think educators now, more than ever before, have that responsibility and calling to do that.

Daniel Herman: Yeah, Danny — it’s funny. I mean, I think my experience at a small, progressive, independent high school in Berkeley is maybe pretty different from Mountain View, where, you know, it’s been a while to see how right-wing-coded so much of AI has become, and the movement against data centers.

There’s a vocal group at my school — I’m not sure if it’s a plurality or a majority, but it’s definitely vocal — that holds that there is simply no ethical use of AI, period. So it’s definitely been, at my school at least, student-led or student-driven.

Alexis Madrigal: A listener writes in to say, “I hope we can keep the focus on the fact that, moral or not, helpful or not, the future of tech or not, AI is absolutely devastating the planet and accelerating climate change at an unheard-of rate.”

We also have another person saying, Martha in Menlo Park: “Reading and thinking is not screen scrolling. Back to books.”

Marjorie writes, “I’ve always been a tech skeptic. What do you do today with elementary kids who have parents who want to limit screen time, favoring more tactile opportunities, and don’t want their kids spending hours each day using screens at school?”

And this is maybe the comment that gets at what I’ve heard a lot from my kids. Ambrosia writes, “My son is a sophomore in high school, a straight-A student, and really is all in and loves school. He’s very frustrated with one of his classes because the teacher is obviously using AI in her lesson plans. She has disclosed that to the students. He has noticed some errors in the lessons and is not only concerned he isn’t learning anything, he is also upset about the double standard. The messaging has been very clear that students can’t use AI because it’s cheating. And now he has a social science class with a curriculum streaked with slop.”

How should high school be handling AI? I love it. We’re coming to the end of the discussion. Unfortunately, we probably could do four and a half hours on this. It’s been so interesting.

We’ve been joined by Daniel Herman, humanities teacher at Maybeck High School. Thank you, Daniel.

Daniel Herman: Thanks for having me.

Alexis Madrigal: You should check out his essays in The Atlantic on this if you’re interested. We’ve been joined by Kip Glazer, principal of Mountain View High School. Thank you so much.

Kip Glazer: Thank you for having me.

Alexis Madrigal: Jeremy Roschelle, executive director of learning sciences research at Digital Promise. Thank you.

Jeremy Roschelle: Thank you!

Alexis Madrigal: Tom Wadbrook, director of technology at Prospect Sierra. Thank you very much. And MacKenzie Price, co-founder of Alpha Schools and Two Hour Learning. Thank you so much for joining us.

MacKenzie Price: Thanks for having me.

Alexis Madrigal: Also, thank you so, so much to all of our callers and listeners. We really appreciate it when you come in on a show like that and bring your best thoughts.

I’m Alexis Madrigal. Stay tuned for another hour of Forum ahead with Mina Kim.

Announcer: Funds for the production of Forum are provided by the Jenerosity Family Fund and the members of KQED.

lower waypoint
next waypoint
Player sponsored by