The TV is in the corridor.
Not broken. Not sold. Just, in the corridor. I moved it while my wife was on holiday. She came back, clocked it straight away, and has asked approximately four times a day since when it's coming back in. My answer every time: it's right there, we can bring it in for a movie night, we just put it back out again after.
The kids were confused for about a day. Now they're not. They play. They argue about other things. It's been, genuinely, transformative.
One episode of Hannah Montana after school becomes two becomes three becomes a negotiation I don't want. The TV being there creates the question. Remove the TV, remove the question. Simple physics.
I've been thinking about this a lot in relation to how I build and how I teach.
The addiction problem.
Building with AI is genuinely addictive. Not in a vague way. In a you'll-lose-a-Saturday way. You're not waiting on a developer. You're not filing a ticket. The tool you always imagined needing is suddenly just... possible, in an afternoon. That's a powerful feeling.
I felt it properly for the first time when I built the flashcard app, the preloaded spaced repetition tool I've written about before (cramly.study). I built it because I thought students needed it. Fine start. But the real moment came when they started using it in class and telling me what they wanted added. It became the thing they actually needed, not just the thing I assumed they needed. They had ownership. That changes everything.
Everything I've built since then has been, if I'm honest, noise. Friction. Another interface to load, another tab to open, another thing for them to learn before they can learn any biology.
That's when I started asking a different question.
AI Slop is optionalNot: what can I build? But: what does this free up?
Splycr (www.splycr.com) started as an attempt to improve on the Kahoot model. It has. Pedagogically, it's considerably more sound: better question design, better guardrails, less chaos for chaos's sake. But I'm not playing it every lesson. I'm playing it instead of Kahoot, occasionally, when the format fits.
What I actually reach for now is simpler: a clean set of short, well-structured questions and answers. I've put real effort into the guardrails. The model I use generates questions that are actually useful rather than just testable. And from that one output, I can make flashcards, Q&A dominoes, bingo, crosswords, word searches. All of it. All in Splycr.
I was first inspired by a book called Friday Afternoon, an A-Level resource built around exactly this kind of thing. Triominoes, dominoes, bingo. Low-tech, tactile, a bit of fun. The kind of activity that works when you're tired and they're tired, and you still want something useful to happen in the room.
Set up stations. Circulate. Talk to students about what they actually understand, not just what they've written down. That's the time I'm buying back.
The actual question.
More and more, I'm asking: how can I use AI to put the devices away?
Not as a gimmick. Not as a rebrand. But genuinely. Using the generation and organisation power of these tools to produce physical, printed, tactile resources that get students talking to each other and get me moving around the room having real conversations about conceptual understanding.
The TV's in the corridor. The question is gone. The evenings are better.
I'm starting to think classrooms work the same way.
The goal was never more technology. It was always more time. To think, to talk, to actually be in the room with the people in front of you.
What low-tech activities are you still reaching for? Always curious what other teachers have laminated into oblivion. Drop me a message.