The Problem
Teachers don't need to be taught the tools. They need a standard to judge them against.
“I'm being asked to supervise students using a technology I don't understand, with no framework for judging right use, and the only guidance I've had came from someone selling me something.”
A summary of what teachers have told me, not a single person's words.
Not what everyone assumes
The tool gap is closing. The judgement gap is not.
The assumption is that teachers do not know how to use AI. That was true in 2023. Free training now solves it and free training is everywhere. If not knowing how to use the tools were the problem, it would already be closing.
What has not been solved, and what almost nobody is teaching, is judgement, the ability to decide whether a given use of AI belongs in a given piece of work, and to defend that decision to a student, a parent or a colleague.
| What free training gives them | What is still missing |
|---|---|
| How to write a prompt | How to tell whether the output is worth anything |
| How to use a specific tool | How to decide whether the tool belongs in this task at all |
| Features and workflows | Where the line sits for this age group, this subject, this assessment |
| Enthusiasm and permission | A defensible basis for saying no to a student |
How it actually shows up
Six situations, all of them weekly.
Every unit of the programme answers at least one of these directly. If a unit answers none of them, it is content rather than remedy.
An essay comes in that reads wrong.
No detector is reliable, and the teacher knows it. So there are two options: accuse a fourteen-year-old on a hunch, or let it go and feel they have failed the other twenty-nine who wrote their own. There is no third option, and nobody has told them which risk they are supposed to take.
They do not know where the line is.
Is using AI to plan an essay cheating? To check grammar? To generate practice questions? To translate a phrase? Colleagues answer differently, so a student is told one thing in English and another in History, and the stricter teacher looks unreasonable rather than principled.
They are using it themselves and are quietly unsure whether they should be.
Marking, report writing, lesson planning, differentiating a worksheet at eleven at night. Almost nobody has been told whether this is acceptable, so they do not mention it, which means the school has no idea how much of its own output is already AI-assisted.
They cannot answer the student.
“Why can't I use it? You use it.” There is no good reply currently available to them, and the absence of one costs authority in the room.
They are afraid of being the one who gets it publicly wrong.
The false accusation that becomes a parent complaint. The colleague who seems to have worked it out. Being the teacher who is visibly behind. This emotional dimension is usually dominant and almost always left out of the analysis.
The only guidance they have had was written by someone selling them something.
Some of the largest free AI courses for teachers are made or funded by AI companies. EdWeek, 2025These examples do not describe all teacher training.
Being straight about the basis
What is evidenced, and what is my framing.
The six situations above are my framing, not research findings. They are consistent with the survey evidence below, and every teacher conversation I have tests them. The survey data has sources; I name them, including where a source is weak or dated.
| Claim | Source |
|---|---|
| 62% of educators call for both continuing professional development and access to approved, safe AI tools; 87% say AI will require some level of upskilling and training; 48% are very concerned about unclear school or national AI policies. | European Commission, School Education Gateway, Survey on artificial intelligence for teaching and learning, 8 April 2026 (1,130 responses from over 50 countries) This survey describes its respondents, not all educators or evidence of improved pupil outcomes. |
| “There was excitement about its potential from many, but there were also doubts and uncertainty, too, and so we needed to work to help upskill and educate all involved” | Clara Hawking, Head of AI, Globeducate, TES, 14 March 2024 |
| “Many schools won't know right now what AI tools staff are using, not because anyone is hiding anything, but because AI arrived quickly and quietly” | Christopher Dunn, Leading Lights, 6 May 2026, a consultant, so weaker evidence |
| In a 2023 UNESCO survey of about 450 schools and universities, fewer than 10% had formal guidance on generative AI. | UNESCO, 2023 This survey describes its respondents in 2023, not all schools worldwide or the current position. |
Two problems, one programme
Two problems, one programme: what it looks like for teachers, and for school leaders.
| The teacher | The Head | |
|---|---|---|
| What the problem is | “I have to judge this and I have no standard to judge against.” | “I am accountable for this and my evidence is a signup sheet.” |
| How often it is felt | Weekly | When something forces the question |
| Emotional register | Exposure, unfairness, fear of getting it wrong in public | Risk, reputation, accountability |
| What AIM Education offers | The units, a standard to judge against | The instrument and the report, a record that survives scrutiny |
Teachers are never assessed by this programme. The instrument is a professional credential and a measure of the programme, never an audit of the individual, no teacher's answers are shown to their school by name.