The Week The Teachers Got

This week, in a public school building in Champaign, or Trenton, or Sioux Falls, a small room of K-12 teachers is spending five days with the AI tools their students have already been using for two years. It is the first cohort of an eleven-million-dollar National Science Foundation grant, awarded to the Computer Science Teachers Association, and running as AI Professional Development Weeks: CS Foundations for Creating with AI.¹ Between June and August, teachers across Illinois, Indiana, Iowa, Minnesota, New Jersey, and South Carolina will each spend one week together, hands on the tool. Four more states will follow next year. Every teacher receives a stipend, both for the week and for the twelve to sixteen hours of ongoing sessions their local CSTA chapter will run through the school year.²

Jake Baskin, the association's executive director, put the reason for the program in a line. AI is arriving in classrooms faster than AI literacy, and that gap is growing.³ The gap he means is not the students' gap. The students have been closing it, unevenly and often unhelpfully, on their phones since the fall of 2022. The gap is the teachers'. Microsoft's June report on AI in schools put a number to it. Fifty-three percent of educators have had no formal training in the tools their students already use every day, and that is the group the CSTA week is trying to reach.⁴

The Sequence Was Wrong

What is quietly striking about the news is the order it corrects. For most of the last two years, the tool arrived in the classroom before the teacher did. Students began using ChatGPT the November of their sophomore year. Districts, understandably slower than the tool's diffusion, spent 2023 and 2024 writing acceptable-use policies and buying detectors. Ohio, only this month, became the first state to make an AI policy a statutory requirement for every district.⁵ The policy came third. The tool was already in the room. The teacher was often the last person in the building to be given time with either.

The CSTA week reverses that sequence, in miniature, for a few thousand people. Before the tool is in a lesson plan, it is in a teacher's hands, for hours. Before the policy is applied to a Tuesday, the teacher has spent a week building something with the model, being frustrated by it, watching it hallucinate, arguing with it about a reading, showing it to a colleague. What comes out of those five days is not a certification. It is fluency. It is a teacher who has done, at her own desk, what she is about to ask her students to do.

What Fluency Alone Cannot Do

The future of education is not built by any tool that arrives in a school. It is built by what happens between the tool and the person who has to teach with it. The weeks this summer are, in a small way, the first serious attempt to design that middle. The intelligence in the room is going to be the teacher's intelligence about the tool, not the tool's intelligence itself. Whether the model hallucinates on a chemistry problem matters, but not more than whether Ms. Delgado in room 214, in the second week of September, notices the hallucination in a paragraph a student turned in and knows how to open the conversation.

What that requires is not only that Ms. Delgado can use the tool. It is that she can see what her students did with it. This is the part that is hard to give a teacher in a week. Fluency with the model, she can build. A record of how her students actually worked with it, she has to be given. The finished essay does not tell her whether the student typed the whole thing, whether she pasted an argument and then reshaped it, whether she asked the model to write in her voice, whether she caught the model in a bad claim and rewrote it, whether she paused for twenty minutes before starting. All of that used to leave marks on paper. It leaves fewer marks now.

The Marks A Classroom Can Bring Back

The point of a well-designed classroom in the AI era is that those marks are visible again. Not surveillance, not detection, not proctoring every essay back into a room. A visible record of the process, so a teacher who has spent a week getting fluent with the tool can then spend her Tuesdays reading her students the way she used to read a paper draft. The revision they kept. The prompt they discarded. The place they got stuck and stayed there for a minute. This is what makes the week the teachers got this summer worth the eleven million. Not only that they came out able to work with the tool. That they came out able to see what a student did with it.

Two years ago, the country handed the machine to students and asked teachers to catch up. This summer, in six school buildings across the country, a small number of teachers are being given the hours they should have had first. It is not a lot of teachers, in a country of three and a half million. But it is a change in sequence. Someone finally noticed that a policy without a fluent teacher, and a fluent teacher without a visible student process, both stop short of the classroom.

What would your Septembers look like if the person doing the seeing had already spent a week doing the making?