Where the Thinking Used to Show

On June 5, NPR and Ipsos released a poll of 545 K-12 teachers taken between late April and early May.¹ The headline numbers are easy to absorb. Three in four teachers say AI's impact on education will outweigh the arrival of the internet or computers.¹ More than half say it is already making it harder for students to learn to think for themselves.² Three in five say they use AI themselves at work.³

The most quietly interesting finding is the one sitting between those. Fifty-seven percent of teachers say AI is now hindering their ability to assess what their students know. Fifty-nine percent say it is fraying the trust between them and their students. Fifty-five percent say their students are using it, mostly, as a way to skip the work.³

Those are not three different findings. They are one finding said three ways. The instrument the teacher has always used to read a child is going out of tune.

The Instrument

The essay, the worksheet, the answer at the bottom of the page. The teacher does not love these for their own sake. She loves them because, for a hundred years, they have been the closest thing to a window into the way a young mind thinks. The teacher who has spent fifteen years reading sixth-grade paragraphs can hear which sentence the writer wrote first, when the second was added the next morning, the moment she was bored and the moment, eight lines later, when she came back. None of that is in the rubric. It is in the teacher.

That instrument is what the poll is describing. The majority of the people who use it for a living are saying it no longer gives them the same reading. The trouble is not, mostly, that AI is in the room. The trouble is that the page no longer shows the work that produced it.

What Christa Saw

Christa Corricelli, a special education math teacher at Saugus Middle High School outside Boston, gave the poll its most precise sentence. "I think students who aren't already intrinsically self-motivated to be critical thinkers, like that top 1% of the class," she said, "I think people who are not already that personality type, we're going to see those critical thinking skills atrophy over time."²

Atrophy is a careful word. It is the slow loss of a faculty the teacher only knew was there because she watched it work in a child for nine months a year. By the time it shows up in the scores, the muscle is already gone.

Corricelli also said she has had no AI training. "I think we're all just kind of trying not to drown with the whole thing."² Only a third of polled teachers said their school had any formal guidance for student use of AI.¹ The largest technological change the field has ever named is happening, by the field's own account, mostly in private, mostly to children, mostly without the adults knowing what to look for.

What the Next Decade Has to Be For

Schools are not going to roll back from generative AI. The same poll found that nearly eight in ten teachers think schools should be teaching students the responsible use of it.¹ The question is not whether children will encounter the tool. They have already encountered it. The question is what the adults in their lives will be able to see when they do.

The decade after this poll will be defined, more than by any policy, by what kind of artifact a piece of student work becomes. The finished essay was useful when it was a faithful record of a mind at work. If it stops being that, something else has to be. The draft she rejected. The eight-minute pause before the breakthrough. The line she added on her own. These are the new shape of the evidence.

This is the slower theory of teaching with AI. It does not ask the teacher to be a detective at the end of the week. It asks the room to make the practice visible while it is happening, and to give the teacher back the reading the poll just said she lost.

Where Trust Lives

Fifty-nine percent of teachers in the survey said AI was fraying the trust between them and their students. Trust between a teacher and a student is not built by an honor code. It is built by being seen. When a student knows her teacher saw the sentence she wrote at 9:42 PM and the one she crossed out at 9:44, the conversation between them changes. It is no longer about whether she did the work. It is about what she was thinking when she did.

This is one of the reasons Koan has been built the way it has been built. Not to replace the essay or to police the chatbot. To restore the part of the reading the teacher used to do without effort and now cannot do at all. The practice itself becomes the artifact.

A profession's most important instrument has gone slightly out of tune. The poll did not say how to retune it. It said only that the people who use it can hear that the sound is wrong. That is the honest first finding of this period.

If the page no longer shows the thinking, what else, in the room, would?