Don't Catch Them, Watch Them

When Miranda Priestly Walks Into Your Course

What we'll do

We'll run a live sample. Miranda Priestly, yes, that Miranda Priestly, will question a fashion student through a design decision. Either a volunteer will take the hot seat, or the whole room will answer together in the chat and I'll relay your replies to Miranda. Then we'll look at real student chat logs (de-identified) and watch that thinking unfold.

What you'll leave with

A prompt framework you can adapt to any persona, using any LLM. It works with a free chatbot, so you can try it the same day. Five pieces make it work:

1. A clear role: who the AI is and who it’s talking to

2. A stern Socratic warning: questions only, no handing over answers

3. One question at a time: to avoid confusion for the student - and so that you see each answer, not a blur

4. A goal and an exit: the decision students are working toward, plus a readiness check to end the chat

5. The full transcript: students submit the whole conversation, which is your window into their thinking

Optional pieces that make it fun and engaging: a name, a defined role, and a personality. You'll also get a fill-in-the-blank template, and a short guide to what to look for when you read the transcripts.

Is this right for your course?

It works best where there are no single right answers: design choices, case analyses, arguments, plans. Because every student's decision is different, every conversation is different, and there's nothing to copy. It's a poor fit for rote memorization, where there's a correct answer to hand over. We'll talk about both.

Want a preview?

Miranda is available for consultations right now. Copy the prompt, paste it into any free chatbot, and tell her about a design you're dreaming up. Fair warning: she will ask why.

Meet Miranda: AI Design Consultation