Who Is Doing the Thinking

Essay 7.1

Lessons from Nintendo: Game-Based Learning

c. 2014–2015 (estimated) · archived July 1, 2018 on Medium

Read archived article · Source details

In the late 1980s, I remember playing the original Super Mario Bros. for hours on end, trying desperately to save the princess. I ducked, dodged, and jumped to complete levels of increasing difficulty, all while learning something new with each new life.

Failure — or in this case dying from an annoying, little cloud-like creature throwing spiky things at me — carried no consequences. I could try as often as I wanted to, with the resolve to figure out the timing of each projectile thrown at me. In a very real sense, I eventually learned how to be successful on my own terms and at my own pace. I learned from failure.

Don’t get me wrong. Mario Brothers may be fun to play, but it has no direct educational value. Today, I’m thrilled with video games that not only motivate and excite, but also teach. More still, I admire how game-based learning champions self-directed learning, with students progressing at their own pace. As one example, DragonBox teaches basic algebraic operations in the same way that one learns to play Mario Brothers. Here too, players gain experience and face greater challenges as they advance through the levels.

Few understand the benefits and potentials of game-based learning as well as Prof. Jordan Shapiro, a leading authority on the subject, who has written extensively about education technology for Forbes magazine.

At the Global Education & Skills Forum in Dubai, he spoke to a packed room about game-based learning — not just as instructional aids, but also as essential tools to fundamentally change how we approach and perceive education. “Imagine classrooms of students around the world intrinsically motivated to learn reading, writing, science, art, mathematics, economics and marketing not because they want a good grade, not because they want a diploma, not because they want a better job, but rather because they want to understand how things work,” he said.

Too often, students are more concerned with the grade than any meaningful and lasting learning. Much of this is our fault as educators, especially in how we demonize failure, pressuring students to master increasingly complex ideas and material in shorter and shorter intervals. Of course, all of this runs counter to how most human beings learn. Deep, lasting mastery takes time. If I were to play Super Mario Bros. again today, some twenty years after my last power-up, I’m sure I would still have a terrific familiarity with the controls — and save at least a few princesses. How many people can say as much about solving a calculus problem, or about explaining the importance of 1812 in our history?

To gain deeper insight into the dynamics of game-based learning, I spoke with Shapiro, who says that game-based learning will lead the way in making failure and asynchronous learning acceptable again. “I wish I was my students’ only teacher,” he says. “I could say, ‘you’re going to write this essay all year until you get it perfect, because this is what I want to teach.’ We don’t go, ‘Oh, well, you can’t learn that. Now, move on to the next level, even though you never learned this level.’”

Along those lines, I remember begging my parents to buy me a Nintendo Game Genie, a cheating device that, among other things, allowed players to skip levels on certain games. All of this didn’t improve my skill, even as I spent days trying to defeat final levels I had no business playing. I grew increasingly frustrated, resorting once again to entering more cheating codes into the Game Genie, such as invulnerability, flight, and enhanced strength. I could only “win” by cheating.

Of course, as a teacher I have a clear no-cheating policy. I take infractions seriously, and I don’t hesitate to submit serious cases to the student honor council. Still, I understand why some students are tempted to cheat. In most cases, they feel helpless in the face of the fast pace of a course, or the rigorous demands of an assignment. In a hyper-competitive academic environment, we have made cheating an alluring option — however risky.

In a game-based learning environment (in contrast to my Nintendo environment), this allure of cheating fades almost completely. According to Shapiro, in the near future, teachers will rely on even more advanced adaptive learning software, so advanced that it isolates precise areas of difficulties.

“Everybody doesn’t have to be on the same pacing because the computer can adjust,” Shapiro says. “Every time you fail, all it does is tell the computer, ‘Oh, we should try a different way of presenting this problem.’”

As a result, Shapiro tells me, video games can isolate distinct learning styles to teach content faster. But there’s still one thing that they can’t do, no matter how advanced technology becomes. “What a video game can’t do is put that content in context faster, in world context,” he says. Only teachers can accomplish this task.

All the same, Shapiro is quick to point out another important fact. “There’s a lot more happening here than just checker games,” he says, adding that the way we think about knowledge and how it’s used is shifting.

Essay 7.2

Technology: The Faustian Bargain

c. 2012 (estimated) · archived June 30, 2018 on Medium

Read archived article · Source details

I sit down with Sean Murphy, the headmaster of Palmer Trinity, the private school where I teach in Palmetto Bay, Florida. We debate the future of K-12 education and what role technology can and should play inside the classroom.

I had moved the student newspaper to an entirely online format. The following year, I purchased a semi-professional digital camcorder, green screen and advanced video editing software for my journalism and yearbook classes.

By my fifth year at Palmer, I have become known as a “techie.” Much of this has to do with the kinds of courses I teach. Still, I get the sense that some of my colleagues have branded me as a rebel, embracing technology and abandoning more traditional, pencil-and-paper pedagogy.

In part, I meet with Murphy to assure him that I have not gone nutty, and that I abhor using technology simply because it’s “cool.”

We agree on everything: technology can definitely enhance the learning process, but too many students are easily distracted by video games, social media, and an array of remote and online applications.

“Technology is a double-edged sword,” Murphy tells me. “On one hand it puts us in touch with each other and with an infinite array of information. But that flood of stimulation doesn’t help us choose the things that make our lives more meaningful or satisfying in the long run. In fact, the sheer volume of stimulation usually becomes overwhelming and disappointing.”

Murphy walks me to his bookshelf, jam-packed with works on education, writing and poetry — reflective of his extremely intelligent, inquisitive, and deep-thinking nature.

He hands me a copy of Neil Postman’s The End of Education: Redefining the Value of School. I was first introduced to this seemingly clairvoyant author, essayist, and social critic at Brandeis University.

“It’s a little dated,” Murphy says. “But I think you might find some useful, interesting stuff in there.”

Several days later, I’m 40 pages into Postman’s work when a certain passage really catches my eye:

“The computer and its associated technologies are awesome additions to a culture, and they are quite capable of altering the psychic, let alone the sleeping, habits of our young,” Postman writes. “But like all important technologies of the past, they are Faustian bargains, giving and taking away, sometimes in equal measure, sometimes more in one way than the other. It is strange — indeed, shocking — that with the twenty-first century so close on our heels, we can still talk of new technologies as if they were unmixed blessings, gifts, as it were, from the gods.”

Postman died in 2003, but his work remains truer than ever.

People often discuss what famous person, dead or alive, they would most like to meet. Postman ranks at the top of my list. I would love to ask him about one-to-one laptop programs and whether in trying to help by placing devices in students’ hands, we are taking away something more important.

Thankfully, Postman’s Technopoly: The Surrender of Culture to Technology, published in 1993, offers a good glimpse of what he might think. In a C-SPAN interview about his work, Postman grabs my attention:

America has developed a new religion, as it were, and the religion is its faith that human progress and technological innovation are the same thing — and that paradise can be achieved through greater and greater commitment to technology.

It’s no secret that I fully support one-to-one programs, and for good reason: Cutting-edge learning software, interactive web platforms for communicating and sharing ideas, and instantaneous information provide endless learning possibilities, which even Postman couldn’t possibly fathom before his death.

All the same, we must heed Postman’s warning. Paradise cannot be achieved by blindly embracing the next new thing, in this case technology. Simply giving students laptops doesn’t mean that the devices will be used correctly, or at all. Postman would certainly agree, and he warns about the dangers of individuals becoming overly enthralled with “crap.”

“One way of looking at the history of the human group is that it has been a continuing struggle against the veneration of crap,” Neil Postman and Charles Weingartner write in Teaching as a Subversive Activity, published in 1969.

If it is integrated effectively, I wouldn’t use such language to describe technology — quite the contrary. Still, I am cautious about excessive and unnecessary use of it in the classroom.

Outside of my journalism classes, I rarely allow my students to use their laptops — only when they need to work on a project as a designated in-class activity. This isn’t because my lessons are boring, or that kids aren’t interested in learning. The temptation to check sports highlights, fashion blogs, and teen “news” sites is just overpowering.

It’s the job of a teacher to show and promote proper use, but herein lies another problem.

Too often, I hear of teachers frustrated with the simplest technological tasks, such as turning on a projector or computer, playing a film or uploading content to a class page. While they struggle, their well-meaning schools encourage even more use of online learning and technology. Frustrated teachers burn time trying to work something that they know little or nothing about. Postman is spot-on: in this environment, no greater progress can be made.

Murphy and I agree that too often teachers view technology as a panacea for all of education’s problems. But I find solace in that Postman admits that in moderation, technology isn’t necessarily bad.

In this book, I mostly emphasize the bad part. But I admit, happily at the beginning of the book, that anyone who looks at technology as an either-or development, that is either as all good or all bad, is making a mistake.

When I was a college student, my journalism professor, Michael J. Socolow, assigned Postman’s Amusing Ourselves to Death.

I glance through my old, worn-out copy, when I stop at a particular highlighted passage:

In courtrooms, classrooms, operating rooms, board rooms, churches, and even airplanes, Americans no longer talk to each other, they entertain each other. They do not exchange ideas; they exchange images. They do not argue with propositions; they argue with good looks, celebrities and commercials.

I speak about Postman’s legacy with my editor, Bruce Musgrave, who points me to another insightful remark by Postman:

Our politics, religion, news, athletics, education, and commerce have been transformed into congenial adjuncts of show business, largely without protest or even much popular notice. The result is that we are a people on the verge of amusing ourselves to death.

Decades before social networking even existed, he predicted the challenges we would face in retaining a sense of what truly matters.

Teachers have an obligation to help students sharpen the good side of the double-edged sword.

Essay 7.3

Artificial Shortcuts, Real Consequences: The Costs of AI “Humanizers” to Learning

April 19, 2024 · Medium

Read original article · Source details

A wise mentor once shared with me that “experience can’t be taught.”

This observation carries significant weight in today’s educational climate, where students face an onslaught of sophisticated marketing campaigns touting platforms that boast of their ability to outsmart AI detection. These ads offer what seems like an easy way out of the challenging yet rewarding experience of learning to think and write.

Among these services, Phrasly.ai emerges as a particularly egregious example. Just consider this transcript of a social media video advertisement for Phrasly.ai, starring what purports to be a high-school student.

I’ve been using this website called Phrasly to help me write my school essays, and I need to put you guys on it. In the past, there’s always been a little bit of a risk with turning in AI-generated content to my teachers because some of them have been able to detect that it was AI-generated. Phrasly’s useful because it can bypass any Turnitin AI detection. With Phrasly, I just copy and paste any AI-generated content in, and in just a few seconds, it will humanize it for me and make it so that no AI detector can detect anything. Thanks to Phrasly, I’ve gotten all of my essays done in almost no time using AI. Now I can put my laptop down and enjoy my Christmas break!

I reject the suggestion that teachers are antagonists for expecting their students to engage earnestly with their work. It promotes the destructive notion that taking shortcuts is not merely permissible but also clever, and that they should at all costs evade educators with the audacity to challenge their students to think, write, and learn.

This brazen promotion of a tool designed to enable student deception is an outrage. I detest how Phrasly and similar platforms peddle not just a product, but also a toxic mentality that devalues the importance of diligence, critical thinking, and personal achievement through mastering challenges.

The sense of powerlessness that educators experience in this David-versus-Goliath scenario is palpable. In our efforts to promote pedagogical integrity, we confront an industry eager to profit by actively encouraging students to cheat.

In response to Phrasly’s advertising, I reached out to their online help desk, seeking to comprehend their justification for the product’s existence. I received a response from somebody named “Mark,” who replied: “Phrasly is dedicated to upholding academic integrity and actively contributes to blog writing for SEO [search engine optimization] purposes. Our company is committed to the democratization of AI, ensuring its accessibility without fear of external limitations. Moreover, with the rising incidence of false positives from AI detectors, our tool serves as an essential resource for those striving to maintain the authenticity of their human-written work.”

This claim that Phrasly protects students from false accusations is farcical. Phrasly prides itself on using “advanced algorithms to modify AI-generated content,” making it undetectable — the core of the issue here.

As we navigate the ever-evolving educational landscape, we must never divorce imparting knowledge from instilling values. Yet the advent of platforms that generate text designed to elude AI detection systems has already shifted my instructional focus. I spend more and more of my time addressing the ethical implications of using such tools rather than delivering content.

In response to this challenge, I have recently implemented strategies aimed at fostering integrity and honesty among students, despite the allure of AI’s seemingly convenient solutions.

For instance, I now introduce discussions that explore the long-term benefits of genuine learning over the expedience of achieving short-term academic gains. By presenting scenarios that demonstrate the value of original thought and the pitfalls of intellectual shortcuts, I encourage students to reflect on the importance of authenticity.

In my discussions with students, I consciously adopt a tone of understanding, acknowledging the pressures they face in an environment rife with shortcuts, such as those offered by platforms like Phrasly. I liken those temptations to low-hanging fruit — seeming within easy reach yet ultimately unfulfilling. My approach is proactive, focused on steering students towards integrity before they succumb to temptation, rather than chastising them post-transgression.

I advocate for transparency and honesty, reassuring students that while submitting genuine work is paramount, the courage to admit a lapse is also commendable. It is essential for them to recognize that acknowledging an error is a testament to their character and reflective capacity. I assure them that such honesty elicits respect and a supportive response, not punitive action. In taking the step to come forward, they not only preserve their integrity, but also embark on a path to rectifying their academic careers.

This is the greatest lesson we can offer: that the journey of learning, with all its inherent struggles and hard-won joys, is itself a profound source of wisdom — one that shapes not just how we think and write, but also how we live and grow as individuals. In embarking on this journey, we embrace the true essence of education.

Essay 7.4

Generative AI Has No Place in My Classroom

August 9, 2025 · Medium

Later version at PBS NewsHour Classroom · Read archived article · Source details

I’ve never been anti-technology. I’ve even written about ways AI can help students when it’s used with care. But my first real encounter with GPT-5 — released on August 7, 2025 — stopped me cold. This is different. This is dangerous. And I’m finally sounding the alarm.

Since ChatGPT first debuted in November 2022, I’ve relied on a system.

Early in the semester, I got to know each student’s unique voice through smaller, in-class assignments. Only later would I assign longer papers. By then, I recognized the rhythm of their sentences, the way they transitioned between ideas, even the little detours they took before circling back. If something came in that didn’t match, I noticed.

That safeguard is gone. The linked papers were produced by GPT-5 for several of my assignments — I annotated and graded them as if the program were an actual student.

GPT-5 can now mimic a student’s writing so precisely that I struggle to tell the difference. Feed it a few past essays, and it will flawlessly reproduce the quirks, pacing, and diction. It can generate a paper in that same voice — complete with accurate citations — polished to look as if it took hours.

Even a hard-to-find PDF or obscure text poses no challenge; upload it, and GPT-5 will read, analyze, and weave it seamlessly into the work. I’m not revealing any trade secrets here. Students know this. Teachers know it — or they should.

While many educators are working hard to integrate generative AI into their classrooms, I’m now convinced that when it comes to teaching writing and critical thinking, AI takes far more than it gives. The exchange isn’t remotely equal. It’s not a Faustian bargain — it’s worse.

What it offers in efficiency, it erodes in depth. Students can lean on it before developing the habits, patience, and skill that real writing demands. The slow, sometimes frustrating process of wrestling with a sentence, choosing the evidence that best supports an argument, or discovering a new insight mid-draft — these are the moments when thinking happens, however time-consuming, inconvenient, and uncomfortable.

When generative AI shortcuts that process, it doesn’t just “assist” — it replaces the very struggle that leads to growth. In my own classroom, that means AI isn’t just a neutral tool — it’s a detriment to helping students become confident writers and critical thinkers.

When calculators first showed up in classrooms in the 1970s, many teachers believed they would destroy math education. Suddenly, students did not have to memorize multiplication tables, slog through long division, or grind out multi-step problems by hand.

The fear was that basic numeracy would collapse. And yes, the disruption was real — tests had to be rewritten, lessons redesigned, and everyone had to figure out when calculators could be used without hollowing out learning.

But here is the key difference: calculators automated only one part of the process — computation.

They did not decide which problem to solve, figure out how to set it up, interpret the answer, or explain why it mattered. The student still had to understand the question, choose the right formula, and apply it correctly.

GPT-5 is an entirely different kind of disruption. It does not just solve a step — it can take over the entire process: brainstorming, outlining, writing, revising, even embedding accurate citations. That is not replacing a tool — that is replacing the thinking itself.

At this point, I’m convinced there’s only one viable solution: more in-class writing. I’m done assigning large take-home papers. The tech is too advanced, too easy to use, and too aggressively marketed. Companies like Phrasly.ai target students directly on social media, openly promising to help them beat AI detectors and fool their teachers. Now, GPT-5 rolls all of that into one seamless package.

AI companies sell this as empowerment. Their ads glamorize “working smarter,” which in this case really means not working at all. They know exactly how to make cutting corners look cool — especially to already overextended teenagers juggling sports, jobs, social lives, and college applications.

Not every student will cheat, but it’s naïve to think more won’t. Expecting students to spend hours wrestling with prose and research at home when this alternative is just a few clicks away is wishful thinking.

Many educators — though in my circle, fewer and fewer — still argue that there is a place for “ethical generative AI use” in the classroom. I used to agree. But when a free or inexpensive tool can fully replace the thinking process, it stops being a support and becomes a substitute brain.

Yes, cheating has always existed — students could buy papers or pay others to do the work. That was bad. But this is worse: now the cheating is democratized, more ubiquitous, and easier than ever.

So, what now?

In my high school history classes, every substantial piece of writing will happen in class, either in old-school blue books or on Digiexam — a secure, locked-down digital platform that lets me monitor work in real time — so I can see the process unfold as it happens. The process will matter as much as the product.

Students will draft in class, brainstorm openly, and revise alongside peers. Big assignments will be broken into smaller, visible stages so I can watch their thinking evolve. I’m also expanding weekly seminars and debates that require students to prove — on the spot — that they’ve engaged deeply with the text. Something will have to give; with more time devoted to writing in class, I may cover less content. But I’d far rather accept that trade-off than watch anyone hand over their intellectual growth to a machine.

The point isn’t to ban tools — it’s to ensure students can think without them. A computer can churn out a paper in seconds, but what matters is whether a student can explain that paper’s ideas in conversation, adapt them in debate, and defend them when challenged.

Because when students hand over the thinking to a machine, the loss isn’t just academic integrity — it’s intellectual independence. We risk raising a generation that trusts algorithms over their own judgment.

And that’s not just bad for education — it’s bad for the future of our world, and for nearly everything else that depends on it.

What I Reconsidered: AI, 2023–2026

Essay 7.5

Don’t Let AI Climb the Hill for Our Kids

January 26, 2026 · Teachers on Fire Magazine (Medium)

Read original article · Source details

The Boston area just received nearly two feet of snow. The timing felt especially fortunate. My family lives directly behind a great public park, and the park includes a terrific hill for sledding. Early this morning, before other footprints appeared, my son and I bundled up and headed out to imprint our fresh tracks.

I did not expect that walk to shape how I think about AI and learning, but it did. Standing at the bottom of the icy hill, I watched my son size it up for the climb to the top and found myself thinking about effort — who expends it, who avoids it, and what disappears when struggle gets outsourced.

“I think it’s very possible,” my son said as he powered up the small but mighty hill. Not too much later, he exclaimed, “Dada, I did it!” His seven-year-old face beamed with joy.

My son’s motivation also explains why I remain cautious about my students using GenAI at all — certainly without my supervision. I want them to know they can achieve their goals on their own, without relying on unnecessary support, especially before they have proven they can do so independently and unaided.

Nassim Nicholas Taleb gave a name to that phenomenon — to my son’s triumph, small as it seemed — terming it “antifragility,” a concept he developed in Antifragile: Things That Gain from Disorder. Taleb argues that certain systems grow stronger through stress and small failures rather than in spite of them.

Jonathan Haidt and Greg Lukianoff later applied similar thinking to childhood and education in The Coddling of the American Mind, warning about shielding young people from difficulty.

Haidt articulated that idea directly to me during our interview in 2019, after the book’s September 2018 release. Reflecting on how children learn, he told me, “There are two major kinds of learning… One is propositions given by adults. The other kind of learning is feedback from experience.”

Haidt emphasized that learning rooted in experience, especially when it involves mistakes or minor pain, sticks faster and runs deeper. “The simplest kinds are times when you get hurt,” he said. “That learning is extremely rapid.”

That framework resurfaced immediately on the hill. Snow had softened the surface. Ice waited underneath. My son climbed, slipped, and slid back down. I could have stepped in. I chose not to. He faced no real danger, and the worst outcome involved another slide or a few bumps. Those bumps mattered. They delivered feedback and demanded adjustment.

My son tried again. He leaned forward. He changed his footing. He surveyed the terrain. After a few false starts, he reached the top. The moment felt commonplace, but it mattered. He proved something to himself. Effort worked. Failure did not end the attempt. Help did not need to arrive immediately.

That moment reminded me of something else Haidt mentioned during our conversation. When I shared concerns about students missing chances to develop resilience — well before ChatGPT launched in the fall of 2022 — he immediately agreed. “Kids are antifragile,” he said. “They need a wide variety of exposure to challenges.”

Small bumps and bruises, physical or emotional, do not reflect harm; they signal learning.

That realization sharpened a concern I witness daily in my classroom in 2026. AI increasingly intervenes before students struggle. Like lifting a child up a hill, it delivers a faster result. AI could not climb for my son, but it could have told him how to avoid the hard parts. In doing so, it would have removed the very struggle that made the moment meaningful.

I notice the same pattern with students. When AI supplies structure, phrasing, or answers too early, learners complete their work without gaining experience. They skip the frustration that forces revision, the uncertainty that builds confidence, and the realization that they can troubleshoot without creating a crisis out of every setback.

To his credit, Haidt warned about this dynamic long before AI entered classrooms at scale. Children, he argued, learn best through feedback loops — thousands of small interactions that teach judgment, resilience, and independence. When those loops disappear in the name of protection, efficiency, or convenience, fragility replaces strength.

Developing antifragility depends on friction. Growth requires stress that challenges without overwhelming. Taleb warned that systems designed to eliminate volatility grow brittle over time. Haidt echoed that warning in education: If adults treat children as fragile, children behave accordingly.

I wonder what happens if, years from now, my son reaches for AI before testing himself. What if convenience becomes the default response to challenge, replacing effort?

The danger of AI does not lie in its existence, which I do not fear, but in the timing of resorting to it. Used too early, AI replaces the work that builds capability. The hill behind my house — and my son trying to climb it — offer a clear reminder.

Growth does not come from perfect solutions, laid out on a silver platter at the click of a button. Growth comes from effort, discomfort, and the willingness to keep climbing.

Essay 7.6

Teaching About AI Is Necessary. Teaching With AI Should Be Optional.

January 20, 2026 · Medium

Read original article · Source details

I’m not anti-AI in education, and I’m not naïve enough to think teachers can just wait for it to go away. AI already reshapes how students find information, complete work, study, and even imagine their futures.

I offer that disclaimer not as the point of this piece, but because even a measured critique of AI these days seems to require one. Ready the trolls!

Here are my real claims: Schools should teach students about AI. Schools should not mandate that every teacher teach with AI. Ignoring AI isn’t a plan, but neither is succumbing to its hype.

Since ChatGPT arrived in November 2022, the loudest version of this debate — especially online and in the so-called “room(s) where it happens,” to borrow a phrase from Hamilton — has largely ignored the space where teachers actually work: the complicated, high-stakes ground zero, inside the classroom.

What keeps me up at night is how often I hear this claim, particularly from people outside daily classroom life: AI literacy should be treated like reading and writing — and therefore integrated across all classrooms.

On the surface, that sounds logical. I get the simple appeal. But when I run that theory through classroom reality — students with varying abilities, limited time, accountability pressures, uneven readiness, and wildly different course demands — it starts to fray.

Not because AI literacy doesn’t matter, but because the phrase “integrated across all classrooms” often smuggles in something else: an assumption that embedding AI tools into instruction should become the default.

I want students to understand what AI is and what it isn’t; why it can confidently invent information; how bias enters outputs; what ethical use demands; and when using it strengthens thinking versus replacing it. None of that feels radical. It feels sensible.

What I’m pushing back on is the growing expectation — sometimes stated outright, sometimes implied — that teachers should embed AI tools into instruction as proof of modernity, relevance, or “future readiness.”

That expectation rarely originates in classrooms. It more often arrives through initiatives, policy language, and people fluent in phrasing that travels well online and to the right ears. Those systems stay tidy. Classrooms are not.

I hear polished phrases like “intentional integration,” “ethical use,” and “future readiness.” They sound good. They fit neatly on slides. They perform well in posts and professional development meetings. But classrooms don’t operate on slogans. They run on time, attention, skill-building, and constant tradeoffs.

Teachers already work to…

• Protect the integrity of assessments
• Rebuild foundational reading and writing skills
• Support students with dramatically different readiness levels
• Manage cognitive load, motivation, and disengagement
• Adapt as expectations keep shifting

AI tools dropped into that environment don’t remain neutral. They reshape what students practice. They alter where they struggle. They change what they skip.

For example, when students depend too heavily on AI for writing, efficiency increases on the surface, but cognitive work often decreases underneath.

Instead of struggling to form ideas, select evidence, and construct sentences, students can move straight to a polished product. The writing looks complete, but the thinking that should have produced it often remains underdeveloped.

This shift adds instructional complexity. Teachers must determine whether missed understanding stems from conceptual confusion, weak skills, or reliance on a tool.

At the same time, evaluating student learning becomes less dependable — not because teachers lack ingenuity, but because confirming authentic work now requires more attention and energy than classrooms realistically allow.

Sometimes using AI makes sense. Sometimes it doesn’t. The problem starts when we pretend there’s no cost — and when we frame adoption as the cost of admission.

A crucial distinction keeps getting blurred, and it matters.

Teaching about AI means helping students understand…

• How responses get generated
• Why outputs can be unreliable
• What these systems routinely get wrong
• How AI shapes society
• Where ethical boundaries belong
• How to verify information and think critically in an AI-saturated world

Teaching with AI means embedding tools into instruction, assignments, assessments, and workflow.

These are different responsibilities. Yet public discourse often treats “not integrating tools” as evidence of being behind — or worse, resistant to progress. That sets up a false choice. Professional judgment still counts.

Every subject is different. Every classroom is different. Every group of students brings dissimilar needs. A journalism class grappling with sourcing and verification faces different risks than a computer science elective.

A ninth-grade history class still developing sentence fluency faces different challenges than a discussion-driven seminar full of confident writers. Treating integration of AI as a universal “must” flattens all those nuances into a single, convenient proposition.

Often the most responsible call is not yet, not for this task, not with these students, or not in this way. That isn’t fear. That’s professional care.

To me, the core problem isn’t individual teachers. The education system responds to certainty more than nuance. Universal claims travel farther than caveats. “All teachers must” spreads more easily than “It depends, and teachers should decide.” But mandates aren’t necessarily wisdom, and I feel fortunate to work at a school that understands the difference and operates accordingly, valuing nuance and amplifying the voices of students, teachers, and the wider community.

In the broad educational landscape, here’s what I want if we’re serious about AI literacy: restraint and humility.

• Teach students about AI before demanding widespread teaching with AI
• Support teacher learning without turning adoption into a loyalty test
• Acknowledge that some learning goals benefit from less AI, not more
• Treat integration as a tool choice, not a moral absolute
• Trust teachers to decide when AI serves instruction — and when it doesn’t

AI exists. It isn’t disappearing. But its mere existence doesn’t always justify its use.

The most important skill we can model for students right now may not be tool fluency. It may be judgment: knowing when a tool supports thinking and when it stunts it.

That judgment should begin with teachers, where it should stay.

Essay 7.7

The Pedagogy of Presence

September 7, 2026 · Medium

Read original article · Source details

That experience changed my classroom. I delayed assigning substantial take-home writing until I could familiarize myself with each student’s unique style. I moved more writing into class, where I could watch them think. I wanted to see the process, not simply collect the final product.

I also added culminating one-on-one conferences, not to “catch” academic dishonesty, though that did happen on occasion, as when I questioned why students cited a paper about World War I while writing about something entirely different in World War II (an AI hallucination).

My goal, which I wrote about in a recent Edutopia post, was to better understand students’ learning processes and offer live, immediate feedback rather than written feedback, which could be misconstrued as “mean” or “aggressive” (a different issue entirely).

I asked Astra to provide exemplars for a take-home essay topic I assigned at the end of last academic year, and I was blown away by the results, which contain (as far as I can tell) accurate footnote citations and clear, analytical writing.

I also instructed Astra not only to cite directly from the resources I provided to students, but also to conduct its own independent research. Of note, I offered Astra no additional instructions or refined commands beyond what I provided to students in the assignment and the rubric.

The biggest stride forward from GPT-5 that I notice, however, is the AI’s ability to embrace nuance. I was stunned, for example, by the first part of the conclusion to its first exemplar response:

Truman’s decision should consequently be criticized without pretending that he possessed a painless alternative or perfect foresight.

Japan’s leaders bore responsibility for aggression and continued resistance; American troops and occupied populations had urgent claims to protection. Those truths justified determined efforts to end the war.

They did not justify treating diplomatic uncertainty as evidence that negotiation was useless, military value as permission to devastate a population, or the first authorization as sufficient warrant for a second attack.

Agree or disagree with the conclusion, this is human-like, nuanced reasoning that makes this history teacher wonder what the technology could do at its highest capability level. My biggest (and growing) concern, though, is how much responsibility students can now delegate.

Generating an answer already lets them bypass the writing itself. An agentic system can take over the rest: finding evidence, organizing the project, producing files, and checking whether the result meets requirements.

But clearly, the biggest question writing teachers have to face now is why it’s even worthwhile to teach writing anymore, when AI can write so well, and with such nuance. I can only answer that writing is thinking, and just because a machine can do something doesn’t mean that it should.

To the critics who say I am ignoring the inevitable, and that I should instead teach my students how to use LLMs and now agentic AI, I can only say that we will certainly discuss the technology, its capabilities, and its implications for, well, just about everything.

My role is not necessarily to equip students with whatever skills employers may covet next — which is an impossible feat anyway — but to help foster inquisitive, independent minds capable of thinking, questioning, and reasoning for themselves, so that however nuanced and intelligent AI becomes, my students will be equipped to question it and push back where appropriate.

I believe that capacity will remain essential.

A note on this collection (2026): I recognize the irony: I used Astra to help assemble this book while arguing that students need protected time to develop the skills it can perform. My Author’s Note explains that distinction.