The key is demonstrating AI as one step in your process — not the main event — so students see it as a useful tool you control, rather than something that runs the show.
The Demonstration Trap
When educators first start showing AI to students live, there’s a common pattern: the AI demo becomes the session. You open Claude, ask it something, read the response, discuss the response, ask a follow-up, and suddenly 20 minutes have passed and the students are watching you have a conversation with a chatbot instead of learning what you came to teach. This is the demonstration trap, and it happens because AI responses are genuinely interesting — they invite curiosity and tangents.
The fix is to treat AI the way a cooking show host treats a blender: you show it doing one specific job, acknowledge it briefly, then move on to the next step. The blender doesn’t become the topic. The dish does.
How to Structure an AI Demo That Stays Contained
Before your session, decide the one thing you want to demonstrate — not three things, one. “I’m going to show you how I use Claude to generate a discussion prompt on the fly” is a contained demo. Set a timer for 90 seconds. Show the prompt, show the output, make one observation about what you notice in the result, and close the tab. That’s it. If students have questions, collect them and address them at the end of the session rather than letting the demo expand mid-stream.
For more involved demos, pre-run the AI interaction and paste the output into a slide or doc beforehand. Students see the result without watching the generation happen in real time — which cuts the waiting and the temptation to go off-script. This works especially well for showing long-form outputs like lesson plans, email drafts, or course outlines.
What This Means for Educators
As a trainer or coach, your credibility comes from your ability to curate and guide — not just generate. When you show students AI in a structured, purposeful way, you’re demonstrating expert judgment: you know what to ask, you know what a good result looks like, and you know when to stop. That’s the model they need to see. An uncontrolled demo teaches the opposite — that AI is unpredictable and hard to manage.
The Simple Rule
One demo per session, 90 seconds maximum, one specific use case. Pre-run anything complex. Keep your hand on the wheel the entire time so students learn to see AI as a tool you use — not one that uses you.