Thoughts and opinions by Brandon Kowallis; editorial work a team effort between Brandon and Claude.
In more than twenty years in public education, I’ve watched the same pattern repeat every time a disruptive technology arrives. The students adopt it immediately. The faculty and administrators rally against it or ignore it. And rather than adapt the system to the technology, we exhaust ourselves trying to force the technology back into a system built for a world that no longer exists.
Artificial intelligence is not the calculator or the internet. It is a tidal wave already reshaping nearly every profession our students are training to enter. Yet on most campuses, far more energy goes toward catching students who use it than toward teaching them how to use it well.
So let me ask the question I think we’re avoiding. Why are students using AI irresponsibly? Because no one is teaching them how to use it responsibly. And why is no one teaching them? Because too much of the faculty’s time and energy is being funneled into complaining about AI, and too little into the hard work of redesigning curriculum, programs, assessments, and expectations around it. It is far easier to write an article about how bad AI is than to rebuild an entire course to help students learn to use it honestly.
If I were king for a day, here are my edicts — but first, a caveat. The best answer usually sits somewhere between the radical and the status quo. What follows will sound radical. Faculty may scoff from deep in the comforting, sometimes blinding weeds of their disciplines; administrators may roll their eyes from the 30,000-foot view, seeing the whole landscape but not what lives in those weeds. Yet soldiers cannot win a battle without seeing what only their commanders see, and commanders cannot read the front without their soldiers. No one solves a new problem by thinking the same old way. So rather than dismiss what you’re about to read, I would invite you to ask where the balance lies — something you can only do after you’ve immersed yourself in the technology, in the trends reshaping industry, and in the actual lives and minds of your students. Three things our system rarely does at the depth it must.
Edict #1: Rethink what we measure.
For more than a century, the well-researched paper, the written exam, and the prescriptive assignment have been our key measures of learning. That usefulness is now collapsing under the weight of new technology. A capable AI can produce the paper, the website mockup, and step-by-step instructions for an assignment from a few lazy prompts — prompts that demand almost no thinking and leave students substantially worse at the skill we’re trying to instill. The answer is not to ban the tool. It is to stop treating the output AI can fake as proof of mastery, and instead measure the human work behind it that AI cannot replicate. Strip away the paper, the assignment, and the exam, and education, at its core, is about making people good at three things: learning, discovering, and communicating what they’ve found. There are countless ways to assess for those if we are willing to rebuild our systems to be more flexible. But that requires hard, and it requires different — two things many are unwilling or unable to be.
Edict #2: Protect the writing that makes us human.
There is writing we do to process our own thoughts, and writing we do to deliver information. The first is where joy lives and where minds both young and old build strong problem-solving muscle. It is deeply personal, and if assessed poorly it can convince a student that because they aren’t a polished writer they must not be smart, creative, or ready for college and a meaningful career. Hand that type of writing to a machine and we grow weaker in nearly every part of life. Educators should get students doing this kind of writing on their own, ideally on paper, as early and as often as possible. But writing is also a delivery tool where conveying the message quickly usually matters more than perfecting the prose. If you can do both that’s great! However, most people lack the time or the skill. AI has both. And does it really matter how the message was built, as long as the reader receives the message the writer intended to convey, in the quickest and easiest way? A college research paper is nothing more than a way to prove a student can research, question, internalize, and integrate knowledge. If AI can now mimic that proof convincingly, it is time to find better ways to teach and require writing — and to let AI become the gift it can be, consolidating findings into a clean draft in minutes so we can spend our hours on the writing and thinking that helps us grow and generate new ideas.
Edict #3: Rethink what we teach.
Reading, writing, and arithmetic suited a world that needed clerks and human calculators. That world is long gone. And because of that, it behoves us to consider a new foundation around one question every student should be able to answer: what is AI good at, and what am I better at? We no longer spend sixteen years turning children into calculators; we moved past that, and today’s college mathematicians reach far beyond what their peers in the early 1900s could. AI accelerates that curve exponentially. Redesign our programs from middle school through PhD and a high schooler could be doing the work doctoral students did ten years ago — gathering data, directing the systems that process it, judging which sources are legitimate, compiling offline information into sets a machine can use, and asking questions no one has asked before. Imagine middle schoolers doing real research in history, economics, and science alongside AI, arriving at college ready to make discoveries instead of repeating them. We spend sixteen years training students to repeat what’s already known and only invite them to discover at the very end. That order is backward, and AI lets us flip it. It also forces an honest admission: AI is changing not just how we teach but what is worth teaching. Will hand-writing code give way to “vibe coding,” where you describe what you want and the system builds it? Most of us hope not and act as though our own discipline is safe — but hope is not a strategy, and the day it gives way to cruel reality is arriving fast. If whole categories of work are being automated, higher ed is obligated to ask which parts of its own curriculum are being automated too.
Edict #4: Redesign the classroom around defense, not delivery.
Picture a course where a student builds a database of their own self-generated writing samples, learns what makes research rigorous and how to tell good information from bad, crafts the prompts and agentic systems that assemble a draft to that standard, gathers offline data the model can’t reach, audits the output against real criteria — then stands in front of the class to present and defend their research, discoveries, and results, corrected in real time. Throughout, the student does the hard human work while the instructor assesses against a clear rubric: Did they use the tool honestly? Were the sources legitimate? Did the process teach them something? And because today’s AI still fabricates, flatters, and launders bad sources into confident prose, students must be taught to hunt those failures — and faculty must assess whether they can. Otherwise we are training a generation to trust the very thing they most need to interrogate. In such a class, students would learn early that shortcuts lead only to embarrassment. I know the idea of embarrassment and failure makes academic leaders shudder. But in a class where failure that leads to discovery becomes the cultural norm, and where students and faculty are encouraged to support one another through that learning journey, might be the most powerful experience we could offer and the very thing that propels even the most fragile students to real success.
Edict #5: Retool ourselves, and flip the classroom.
None of this will be easy, and we can’t afford to dilly-dally. If a high schooler can generate a research-level paper with little effort, the skill we must teach is no longer how to produce that paper but how to understand and defend what was produced. That means assessing understanding in real time instead of handing out a final exam — initially far more work for faculty, and significantly less time at the lectern that many of us love. It means faculty must become genuine, current experts who can correct a student on the fly and who understand these tools more deeply than their students do. Years ago I flipped my own classrooms. It was a mountain of work up front, but in the end it made the class far more fun and engaging, made it easy to fold in new developments from my industry, and freed me to stay current on the latest trends. In the age of AI, the flipped classroom stops being a novelty and becomes the norm, where students learn outside class and come to class to be challenged, corrected, and guided. That is harder for everyone. It is also better for everyone.
The central purpose of education is to prepare people for life and for work. Both are being remade by the tsunami of artificial intelligence. We can keep spending our energy trying to stop students from touching it, or we can do the harder and more honorable thing and rebuild what and how we teach so that when the wave breaks over our students, they already know how to swim.
If you’re an educator, administrator, or student finding creative ways to work with AI instead of fighting it, I want to hear from you. I feature people doing exactly that on my podcast, The White Squall Podcast. You can reach out to me right here on my website — you might be my next guest.
In full transparency: I wrote this essay with AI, and I’m not going to whisper about it. Every idea, argument, and opinion here is mine, argued and revised over many rounds; Claude served as a thinking and editing partner — precisely the way I’m arguing students should learn to work. That is the whole point. Used lazily, AI produces generic sludge any careful reader can smell in a sentence, and dismiss just as fast. Used well, it sharpens your thinking and multiplies what you can do. I’ve published a companion piece here on my website that walks through exactly how this article was made, so you can see the difference for yourself. I’d ask only this: don’t dismiss an idea because AI touched it. Judge it by whether it’s true.
