AI Summer School

I lost the plot more than once this year, but a few things held

One of the most unique things about teaching is that we exist within discrete units of time. Our year is the academic calendar, not the seasonal one, and each carries its own distinct themes, accomplishments, and struggles. 2025–2026 was a strange one for me – in some ways one of the more difficult of my thirty-one years, in others one of the most rewarding. I wrote a couple of weeks ago about a kind of AI exhaustion that had settled over me. It still lingers.

Still, maybe because I can practically reach out and taste summer right now, with the daily grind of classes receding, I'm already feeling the pull to return to the AI question in earnest (that didn't take long!). This summer I plan to approach it from three angles: the practical, the pedagogical, and, increasingly, the philosophical.¹

The Practical: What’s Hype and What’s Real?

By practical I mean keeping up – not only with the models and the updates, though certainly those - but with the larger issues lurking in the background. I follow the news closely; I can’t help it, it’s my nature. During the school year it’s easy to get lost in the latest shiny thing, but summer is the chance to step back, survey the landscape, and also to play. Playing with AI has only gotten more interesting as the possibilities keep expanding.

Much of the AI discourse² continues to traffic in sweeping claims that bear little connection to my reality on the ground, fueled by the news of the day and usually folded by each commentator into whatever they already believe. Little of it is grounded in what AI actually does - much of the pontificating marinates in abstraction.³ AI is not all one thing, and yet so much of the writing about it draws confident conclusions and makes definitive predictions that either avoid nuance or refuse to engage a single counterargument. This applies to all sides.

The other end of the AI discussion occupies a more granular and niche slice of the conversation - prompting guides, productivity hacks, and inside baseball secret AI strategies offering to give users an edge in … something. Many of these posts fixate on models, examples, jargon, and workflows. This kind of writing is not informative beyond its narrow purpose and simply assumes AI is mostly beneficial. The loudest of the AI hype crowd online are often these writers and their followers.

The two camps rarely talk: the skeptics don’t engage the boosters, and the boosters write off the skeptics as out of touch. Almost everyone is talking past each other. This has more or less been the case for the better part of two years.

My third way has been to simply move forward using my own experience as the best test case. I’ve seen the studies and try to read all viewpoints. I have some strong opinions but I reserve the right to change my mind, which I’ve done on more than one occasion. I have a list of tools I want to try this summer, mostly to see for myself what’s overhyped and what isn’t. I’ve had success building some simple static sites this year, but I know I’ve barely scratched the surface and I want to explore more.

This is a hobby, but AI has allowed me to do things I’ve wanted to do for years: my current goal is to architect full-fledged online academic courses based entirely on my own specifications, the kind of project that would not have been possible a few years ago. I’ve already started a total redesign of my Government and Politics course I’ll be teaching in the fall, inserting House and Senate trackers for the midterms, a navigable glossary of political terminology, and all eight major units embedded on a single site. I’m unsure whether any of it will justify the hours I’m going to invest, but I’ve never found a better way to learn something than to just do it. The beauty of AI is that if it’s not working or you don’t understand something, you can ask.

It’s also why I think a lot of the negative coverage around AI and education right now is aimed at the wrong target. Much of the commentary is still fixated on the student writing problem – who is using AI, how to catch it, or whether to allow it at all. That concern is legitimate, but I think it’s trailing the reality of where things are headed. Students outsourcing their writing won’t stop being a problem, and in many ways it’s already changed many students’ behavior. But using AI as a default search engine or a writing companion is not where the power is going to lie, and my hunch is that a year from now the use cases that matter will look more sophisticated than the ones we’re arguing about today. I’ll be curious if we’re still litigating the essay with the same intensity twelve months from now.

Pedagogy: Where Does AI Actually Fit?

This is the critical question that matters most for educators. There’s no question AI went through a negative cycle this year – backlashes to backlashes and visceral reactions to screens in general, especially as AI got more powerful and intertwined with everyday life. That makes the environment for talking about it in schools more fraught than ever, with contradictory and often incendiary information competing for attention at exactly the moment when the decision-makers need more clarity, not less.

In a year of conversations with school leaders, a pattern has emerged. More of them realize they need to take some kind of principled stance beyond just policies, because it’s clear that the issue is not going to disappear. But centering a school’s pedagogical questions on AI in particular, I’m increasingly convinced, is the wrong strategy. The schools that thrive moving forward will be the ones with a clear sense of mission and the flexibility to meet the pedagogical issues raised by AI as they continue to evolve. K-12 schools that go all in right now without careful planning, as well as schools that refuse to engage at all, will find themselves either reinventing the wheel or constantly playing catch-up.

So what does thoughtful discussion look like? I’m more and more of the view that it mostly looks towards practices that worked before ChatGPT launched – prioritizing student engagement, attention and time on task, scaffolded learning - the same things skilled teachers have done for decades. Teaching in the age of AI does not require a new kind of teaching - at least, inside secondary classrooms. Anyone who has stayed in this profession long enough is used to adapting to new environments. These techniques can be done with or without technology. Last August I wrote about my AI-Aware Strategy for the Year Ahead, and the honest truth is that centering AI as the main driver in the conversation was the wrong move. What I want this fall is simpler – not an AI-Aware Strategy, but simply an Effective Teaching Strategy, where AI just happens to be something we all have access to.

The Homework Question

The real AI disruption over the past two years is the impact it’s had on the work we assign outside the classroom, and that’s where serious rethinking needs to happen. Many teachers have already started. The uncomfortable truth is that the focus on “learning” – which should have been the priority all along – has drifted further and further from daily practice. That sounds counterintuitive, but for a lot of reasons we stopped making actual understanding the main goal of school a long time ago, and AI is now forcing that conversation at a higher level than we’ve had it in years. That should be a good thing in the long run.

A real and growing number of students are already functionally dependent on AI for much of their written work, and that is only going to grow. A second, likely much smaller cohort, will continue to deliberately avoid it – these are the ones who fell in love with reading and with the slow work of writing, either on their own or under the tutelage of an inspiring teacher. Ideally, we can maintain and even increase the size of this group in the coming years.

But the largest group by far is the middle: students who haven’t thought much about any of this, don’t especially care about the AI-and-writing angst, and are mostly trying to get through the day as best they can. They reach for AI in some moments and not others and don’t think very hard about it either way. In every one of these cases, AI has changed the calculus for how students approach a given task, and no AI detector is going to fix it. This is the group that needs the most attention.

None of that means we should stop asking students to do work outside of class. It does mean that any of that work, if we intend to assess it in a high-stakes way, carries an authenticity problem that is not going away. Harsh penalties may deter the worst offenders, but given the infinite number of ways a student can use AI in the context of an assignment, committing finite resources – in time, energy, and trust – primarily to enforcement feels like a losing proposition in the long run.

This is also why AI is both the cause of and a potential solution to the problem of out-of-class work – the cause, because students will keep outsourcing their thinking if we don’t re-examine the tasks we give them; a possible solution, because we can use AI to help us rethink those tasks. I wrote about this previously and I still believe it: in the absence of access to an expert curricular designer, AI can be used to help a thoughtful teacher analyze what may be working in their course and what isn’t. Running your own curriculum through that kind of audit is a solid place to start, and something I’ll do more of this summer.

That’s for the near-term. A more hopeful future, perhaps hard to envision in the current climate, is one where students come to see AI as something that can augment their own abilities rather than replace them, and where school teaches the core skills first, so that the tools are actually useful once students reach for them. I’m not sure how we get there, but as I watch another group of seniors head off to college, I want to believe that we can.

Student Saviors

What got me through the year, in the end, was the reason I’ve remained in the classroom for over three decades – the kids. Time and time again, my students saved me from myself. Real, genuine human connection is the heart and soul of any classroom. We learned this instinctively during the pandemic. Without hearing the enthusiasm in a student’s question, their frustration when they still don’t understand, and their joy when something actually clicks, none of this really matters. No chatbot can read those cues (despite what the new voice models are claiming). Even if they could, it’s clear that’s not what teenagers want.

My students this year were curious, hardworking, and genuinely wanted to learn. They asked tough questions and put in extraordinary effort, and it is clearer to me than ever that what they want most is a caring adult they respect at the front of the room. They prefer the kind of direct attention that only we can give them. My AI Club lunches, Saturday debate tournaments, and Ethics Bowl run to the finals were all highlights of my school year.

Where AI can help is in the hours we aren’t there – when students are guided and motivated to use it to deepen their understanding, apply new research skills, or get feedback on a topic beyond their teacher’s expertise, ideally after they’ve built the skills that let them use it well. The tyranny of assessment hovers over all of this, distorting and too often undermining the very learning it is meant to measure. In an ideal world we help students reach their full potential with these tools if they want them, once that foundation is in place. This is, of course, easier said than done.

What I don't think serves anyone is pretending AI hasn't already worked its way into so much of student life. Pretending the situation doesn’t exist, or trying to ban and/or punish our way out of it, only guarantees that students will use AI in the shadows. My message in the K-12 setting is to center the learning first, foregrounding student needs with the understanding that many of them are already turning to AI. A former colleague used to say that no technology tool was inherently good or bad - user intent coupled with the potential for positive learning outcomes made all the difference. It would be a major breakthrough if we reach the stage of discussing AI in school settings where those considerations drive the conversation.

---

Connect With Me

Beyond this newsletter, I work directly with schools, educators, and other institutions about pedagogical questions raised by AI. Take a look at my website and reach out - I’d love to hear what you’re working on.

1. In another post, I’ll explore the more philosophical goals I have for the summer by revisiting the texts I read for a preceptorial on Education and Pedagogy a few years back. Right now, I want to focus on the work my seniors have done this year and celebrate with them this week as they head off to their next chapter after Saturday’s graduation.

2. By discourse, I mean not only here on Substack, but also in the larger national conversation - the comment sections of mainstream and legacy media and the constant tit-for-tat articles that seem to pop up everywhere. It’s fairly clear at this point that writing about AI generates clicks, and the algorithms are feeding each of us our preferred versions of the arguments.

3. While by no means do people need to stay on top of all the most recent news, if you want to argue forcefully on issues surrounding AI, you should at least try out the latest models and capabilities, if only to get a sense of how far they have come in the past couple of years.