What teachers actually want from school AI training

After running dozens of staff PD sessions on AI in schools, the gap between what gets booked and what gets used is wide. The booking is often a vague brief. "We'd like a session on AI for our teachers." The using is what happens, or doesn't happen, three weeks later when the same teachers are back in their classrooms.

The teachers who get the most out of these sessions ask remarkably similar questions, regardless of the school. The questions are not the ones that show up in the marketing material for AI tools.

The first question. Show me what this looks like for the kind of work I actually do. Not a generic case study. A genuinely realistic example, in the actual subject area I teach, with the constraints I actually work under. Year 8 English teachers need to see Year 8 English work. Year 11 chemistry teachers need to see Year 11 chemistry work. The PD session that shows a beautifully crafted lesson plan for a generic "Year 9 humanities" topic gets polite applause and then forgotten.

The second question. What is the smallest possible thing I could start using this for tomorrow? Most teachers don't have the bandwidth to redesign their pedagogy mid-year. They have the bandwidth to try one small thing, in one specific context, and see whether it saves them time. The PD session that gives them one small win, with a clear "do exactly this on Monday" prescription, lands. The session that gives them ten possibilities tends not to.

The third question. What are the things you've tried that didn't work? This is the question I get asked privately, after the formal session is over, when the teachers feel safe enough to ask the real question. Most of them already know AI has limitations. They want to hear them named explicitly, by someone who has actually used the tools in a classroom, so they don't have to learn each one the hard way.

The fourth question. How do I tell the difference between a student using AI thoughtfully and a student using AI to avoid thinking? This question has barely any presence in most published AI training. It has a major presence in the actual teaching work. The teachers who are best at this don't have a magic detection trick. They have a relational practice of knowing their students well enough that the change in voice is visible. PD that respects this is welcome. PD that promises a software detection silver bullet is met with appropriate scepticism.

The fifth question. Where do I draw the line, and how do I explain it to my students and my parents? This is the conversation teachers find hardest, and the conversation most school AI training avoids. It cannot be reduced to a worksheet. It can be modelled. The most useful sessions I have run gave teachers a script they could adapt, in their own voice, for the conversation with a Year 9 student who has just submitted clearly AI-generated work and is now in the room with their parent. Five minutes of role-play work. Hours of usable practice afterwards.

If you are commissioning AI PD for your staff this year, my suggestion would be to brief the trainer not on what they should cover but on which of these questions they are equipped to answer. The session that answers the questions teachers are actually carrying does more in 90 minutes than the session that covers the syllabus does in a full day.