What the Trust Was Made Of

On Friday, NPR and Ipsos released a nationally representative poll of 545 K-12 teachers.¹ The headline number, the one most of the radio cuts led with, was that nearly three in four believe AI will mean more for education than the internet or computers ever did.¹ Fifty-four percent said AI is making it harder for their students to learn to think for themselves.¹ The number underneath was quieter and, in some ways, heavier. Nearly six in ten teachers said AI is eroding the trust between them and their students.¹ Mallory Newall of Ipsos, presenting the findings, called the trust erosion "one of the biggest red flags in the data."¹

That last sentence is the one to sit with. Critical thinking is the kind of worry that drifts into op-eds and op-ed responses, that gets argued for years before anyone agrees what to measure. Trust between a teacher and a student is more elemental. It is the thing the entire classroom rests on. When it breaks, the rest of the architecture has nothing to stand on.

The Quiet Suspicion

Christa Corricelli, a special education teacher at Saugus Middle/High School outside Boston, gave the poll the most useful sentence in it.¹ She said teachers have become "much more suspect" of anything students do outside the classroom.¹ The next sentence followed quietly. "Well, we can't do it this way because they're just going to use AI."¹

Read that twice. A teacher is describing a pedagogical choice being narrowed not by a curriculum decision, not by a school board policy, not by a research finding, but by a quiet suspicion. The novel is no longer assigned, because the response paper will not be the student's. The essay is no longer due Thursday, because Thursday morning's draft will arrive from a chatbot at 11:47 the night before. The conversation in the room about what the student is reading has thinned because the teacher and the student both know the work outside the room has become unverifiable.

Four in ten teachers in the poll said they have responded by requiring more assignments by hand.¹ Four in ten said they have moved more of the work back into class.¹ These are reasonable adjustments. They are also, plainly, a defensive crouch. A profession is rolling the seven hours of school back inside the building because the eight hours outside have stopped being something it can trust. That is a coping strategy, not a future.

What the Trust Used to Rest On

The careful question is what the trust used to be made of. Before this past year, when a teacher trusted a student's work, she was trusting a chain of small assumptions. That the paragraph in front of her had passed through the student's mind. That the second paragraph had been harder than the first. That the conclusion had taken twenty minutes longer than the introduction. That the student had been the writer.

None of that was named at the time, because it did not need to be. The chain was held together by a single condition. The work was, broadly, the student's. The teacher did not have a window into the writing process, but the writing process was inferable. The grade in the margin was a kind of trust receipt.

That receipt has stopped being legible. A finished essay no longer tells the teacher anything trustworthy about the path the student took to get there. The work has become opaque at exactly the place a teacher needs to see into. That is, in a sentence, what the six in ten are reporting.

The Trust That Could Be Built Next

A future of classroom trust does not, mostly, look like better detectors. The detector arms race has already run its course. It does not look, either, like a return to all work happening on paper inside the room. That is the four-in-ten coping strategy, and it scales until it does not. The cohorts of students who write only by hand in supervised classrooms will eventually graduate into a world where almost everything is written into a machine. The pretending will have to end at the diploma.

The future of classroom trust looks more like the chain of assumptions made visible again, in a different medium. A record of the actual work. The first sentence the student typed. The pause at minute four. The revision to the second paragraph at minute eleven. The full-paragraph deletion at minute fourteen. The breakthrough sentence at minute nineteen. The help from Aidan or from anyone else, time-stamped and labeled, in the same record as everything the student did before and after the help arrived. Not surveillance. Provenance. The same kind of trust receipt the old way produced, regenerated in a form that survives the new conditions.

This is what we are quietly building at Koan. The point of preserving the trail of a student's work is not to catch her. It is to give the teacher back what was there before, when she could look at a draft and know, without having to ask, that the words on the page had come through the student's mind. The trust between teacher and student was always built on the visibility of the work. The visibility just became, briefly, harder to see.

The Question Worth Sitting With

The six in ten is not, finally, a finding about a tool. It is a finding about a relationship that has lost its evidentiary base. The schools that get this decade right will be the ones that rebuild the evidentiary base, not by banning the new tools or by pretending to detect them, but by making the act of learning legible again. Including, honestly, the parts of it where the student leaned on a model and the parts where she did not.

The poll is the field's quiet way of asking a question it has not yet answered out loud. What were we trusting, when we said we trusted them? The honest answer is that we were trusting the visibility of their thinking. The work to be done now is to give that visibility back.