The Training Doubled

Education Week posted a quiet piece this week with a headline that did the heavy lifting on its own. "More Schools Are Providing AI Training for Teachers. Is It Any Good?"¹ The first half of the sentence is the optimistic finding. The second half is the actual story.

The numbers are real. As of March, fifty-eight percent of K-12 teachers told the EdWeek Research Center they had received at least one professional development session on AI. Two years ago, the figure was twenty-nine percent. A clean doubling in twenty-four months.¹ Of the six hundred and fifty-one teachers surveyed, most said they were eager to learn more. Only thirteen percent said they had no interest at all.¹

But the headline keeps poking at something underneath the doubling. Whether the training is any good is not really a question about content. It is a question about purpose. What is the training for.

The Average Hides the Gap

A RAND brief drawn from the American School District Panel found that by the fall of 2024, sixty-seven percent of low-poverty districts had trained teachers on AI. In high-poverty districts, the figure was thirty-nine percent. By the start of the current school year, RAND projects nearly every low-poverty district will have trained teachers, while only six in ten high-poverty districts will have.²

The doubling is uneven. The schools where students are already most likely to encounter AI well are also the schools where teachers were prepared for it. The schools where the cost of getting it wrong is highest are the schools where teachers are most likely to be improvising.

What the Training Is Mostly For

The same researchers asked superintendents what the training was meant to accomplish. The answer that came back, again and again, was close to "calm the room." Of fourteen district leaders interviewed, thirteen described training as a response to teacher fear. Worry about cheating. Worry about replacement. Worry about a technology slipping under the classroom door before anyone had explained what it was.²

A teacher afraid of a tool will not use it, so reducing fear is an honorable first step. But it is a first step. The gap between training designed to lower a teacher's anxiety and training designed to deepen her practice is the difference between a session and a skill.

Two Kinds of Sessions

There are roughly two kinds of AI training showing up on district calendars this year.

One kind teaches the teacher to operate the tool. How to write a prompt. How to generate a rubric. How to scaffold a worksheet. The teacher leaves with new shortcuts and her workflow gets shorter. This is useful. It is also the kind of training most easily confused with progress, because it produces visible artifacts at the end of an afternoon.

The other kind teaches her to read the student differently because the tool is in the room. Her student's work is no longer a single finished essay handed in on Friday. It is a sequence. The drafts she kept and discarded. The questions she asked the tutor. The pause before the third paragraph. The teacher leaves with a different kind of skill. She knows what to look for. She can sit next to a sixth grader on Tuesday and notice that the comma is what keeps stopping him, not because she saw it on a worksheet, but because the record made the moment visible.

A district can run both kinds of session and report the same number. From the outside they look identical. From inside a classroom, they are not.

The Money Coming In

This summer the Computer Science Teachers Association will run an eleven-million-dollar initiative funded by the National Science Foundation, training roughly three thousand K-12 teachers across six states.³⁴ Big tech is running parallel pushes worth tens of millions more. The professional development market is being flooded. The question of what the training is for is about to be answered by whoever shows up first with curriculum.

The schools that will get the most from this moment are the ones that decided in advance what they wanted teachers to be able to do. Not what they wanted teachers to be able to use, but what they wanted teachers to be able to see.

A Useful Afternoon

An honest session might begin not with a tool but with a question. Place two student writing samples on the table. One was produced in a single sitting. One was produced over six drafts, with three rounds of feedback from a tutor and two pauses long enough to suggest the student was almost ready to quit. Ask the teacher which student she would teach differently next week. Then show her how AI, used carefully, can keep that second record alive instead of erasing it.

The training doubled this year. The classroom around it has not yet doubled in its ability to receive the time and attention that good training is meant to return.

If your district is sending teachers to an AI training this summer, what should they be able to do for a student on the Tuesday after they get back?