I’m just going to go ahead and answer that right away. The answer is yes. I probably am as well.
I was just at a conference where vendors were selling courses, AI applications to help teachers differentiate, customize and personalize lesson plans. Instructional coaches demoing how to get a chatbot to write a quiz. Breakout sessions on using AI to align standards. You get the idea. It’s all the same stuff we’ve been seeing for two years now.
And I sat there thinking.
Not because those things don’t matter. They do. Lesson differentiation matters. Standards alignment matters. But if you think clicking a button and getting a lesson plan is what AI can do for education, you are barely scratching the surface. That’s the equivalent of buying a printing press and using it to make one flyer. (I took a graphics class in high school and we made flyers and due to budget constraints were limited in how many copies we could make).
The Small Version
Here’s what I see most educators being trained to do with AI right now: individual tasks. One lesson at a time. One quiz at a time. One rubric at a time. Generate, copy, paste, move on.
And the trainings themselves? An instructional coach gets sent to a conference, attends some AI sessions, learns what they can, and comes back to train staff in an after-school workshop or on a PD day. Those trainings are still largely inadequate because they don’t address bias and sycophancy, and the complexity around how these tools actually work. But that’s kind of how PD goes in lots of districts in my experience.
So we have teachers generating individual artifacts with AI tools and thinking that’s the move, when it’s only a starting point.
The Big Version
For context, I’ve spent 26 years building curriculum. I’ve done big instructional design projects. I’ve written full courses from scratch. I know what goes into making something that actually works for learners: content that is accessible, inclusive, engaging, appropriately scaffolded, and built with the kind of care that students can feel when they interact with it.
Twelve years ago, I wrote a year-long course. It took me a year and over 800 hours outside my teaching responsibilities. The burnout was significant.
Two years ago, I was tasked with using AI to build two year-long courses. I did it, working half-time on the project over an entire school year. It was faster, but it was still over 350 hours of tedious, manual work per course. I told people I’d never do it again. Again, burnout.
I have a new reason for why I won’t do that again. I don’t have to.
Over the last few weeks, I’ve been bringing together everything I know about course development, instructional design, and what actually works for students with all different kinds of challenges, and I built a framework: a system that reflects how I think about building courses and what I believe needs to be in them.
I gave my AI agent access to two different LMS platforms so it had the permissions it needed (don’t do this on production sites with student data). I loaded my framework. And then I let it work.
It wasn’t long before I had a complete semester course. A whole course, structured, sequenced, built with the principles that matter to me, and ready for my review and revision.
12 years ago: 800+ hours of work (with burnout)
2 years ago: 350+ hours of work (with burnout)
2 weeks ago: a lot less (no burnout).
The final product was better than what came before it. Available for updates, revisions, all kinds of improvements. And it was fun. And I’m getting better at it. Finding more efficiencies. I can write a course in a few hours now and all it took was a lifetime of experience.
Who Actually Built the Course
I know how that reads. It reads like AI built a course in five hours. It didn’t. I built a course in five hours. The AI was the tool.
If you handed that same AI the same LMS access without the framework I built, you know what you’d get? The default version. And the default version sucks. It has always sucked. It sucked when vendors built it. It sucked when LMS templates offered it. And it sucks when AI generates it without someone who knows better sitting in the driver’s seat.
The AI doesn’t know what a theoretically perfect course needs to be when you are working with real students. That’s what I brought to the development. A career’s worth of mistakes, collected experiences, insights, care. Knowing what student’s desire paths are, how hard I can push on them. It was decades of revision and revision and revision. It’s my creativity, my passion for the process, my humor, my attention to detail… Anyway, you get the point. AI doesn’t have any of that.
I do. That’s the difference.
The AI was doing work I’ve always done, at a speed that was never possible before, because I told it exactly what to do and how to do it. Sometimes it couldn’t then I told it how it could. Every decision in that framework came from years of building things, getting them wrong, rebuilding them, learning what works and what doesn’t, and developing a set of principles that I trust. I also know a bit about how these bias engines work and context windows and available tokens and prompting strategies, and, and and. It’s work, but it’s also worth it.
The Gap
Here’s what frustrates me, the conversation is stuck on task-level AI use. How to make handouts. Writing quizzes. How to differentiate a reading passage. And those are fine, genuinely fine, as starting points. They’re part of the work.
But few are talking about what happens when you move from tasks to systems. When you stop using AI to generate one artifact at a time and start using it to build coherent, principled, complete learning experiences. That’s a different conversation entirely, and it requires a different kind of expertise.
The tools are there. What’s missing is knowledge. Most people using AI for education don’t have the background to manipulate their AI tools or to tell the AI what good looks like. So they get whatever the AI thinks good looks like, which is some predictable blend of everything it was trained on. That’s not good enough for our students. It never has been.
So, Thinking Too Small
We are in a moment where it is genuinely possible to rethink how we build learning experiences from the ground up. Not just faster versions of what we’ve always done, but fundamentally better ones. The tools can do a lot when the person guiding them knows what good work looks like.
I will never believe that AI can do it alone. It absolutely cannot. The people who know what good education looks like can now make change at a scale and speed that was never available to us before.
That’s worth paying attention to.
I’m available for side projects. 😉