Ninety To Eight

On July 21, at its annual conference in Louisville, the learning management company Instructure released a survey that put two numbers next to each other.¹ Ninety percent of higher education students said they use AI in class at least occasionally. Eight percent of K-12 educators described their AI training as comprehensive.²

The distance between those two numbers is not, in the end, a training gap. It is the shape of the room the school year is about to open into.

What The Survey Asked

Instructure fielded three parallel surveys through SurveyMonkey Audience between June 12 and 15, 2026. Four hundred ten current educators across K-12 and higher education. Three hundred twelve current or recent higher education students who graduated after January 2023. Four hundred three parents and guardians of K-12 students. One thousand one hundred twenty five people in total.²

The findings will not surprise anyone who has been near a classroom this year. Ninety percent of higher education students, sixty eight percent of K-12 educators, and sixty one percent of higher education educators reported using AI at least occasionally. Seventy three percent of parents said their children used it.¹ Then the second, quieter finding. Forty five percent of K-12 educators, and forty one percent of higher education educators, reported no formal training on AI at all. Eight percent of the K-12 group described their training as comprehensive. K-12 educators rated their own confidence with AI at a mean of 2.53 out of five.²

What the four groups agreed on is the part worth pausing on. Educators, students, parents, and higher education faculty all named the same three concerns, in the same order. Overreliance on AI. Loss of critical thinking and deep learning. Academic integrity.³

The Ratio That Started The Year

Ninety and eight. It is worth sitting with the mismatch. Nine out of ten students walking into a fall classroom will already have their own working relationship with the tool. Fewer than one in ten of the teachers standing at the board will have what they themselves call comprehensive preparation for what those students are doing.

The number that gets pulled into the headline is ninety. The number that will shape the fall is eight.

What The Training Gap Actually Is

It is tempting to read a chart like this and answer it with more professional development days in August. That is not wrong. It is also not enough. The reason forty five percent of teachers report no training is not, primarily, a scheduling failure.⁴ It is that nobody has yet settled the question of what the training should teach.

Every teacher who has quietly opened a chatbot on her prep period this summer has learned the same thing. The tool can generate. The tool can summarize. The tool can grade. What she cannot yet see, and what no vendor training will tell her, is what her twenty eight students actually do with it in the twenty minutes between the bell and the deadline. The training gap is the symptom. The evidence gap is the cause.

What The Three Concerns Have In Common

Look at the three shared worries again. Overreliance. Loss of critical thinking. Academic integrity. All three are about the same missing thing. They are about the record of the work.

Overreliance means: I cannot tell if she leaned on the tool for the ideas or only for the phrasing. Loss of critical thinking means: I do not know what would have happened in her head if the tool had not been there. Academic integrity means: I cannot reconstruct the process from the finished page. What the four groups are naming, in three different vocabularies, is that the finished essay is no longer enough to see the student through. It was already thin evidence for a teacher of twenty eight. The tool has just made that thinness impossible to ignore.³

The Layer That Is Actually Missing

At Koan we work on the layer under the finished page. The pause before the second paragraph. The sentence deleted and rewritten in different words. The moment a student pasted something in, sat with it, and translated it back into her own language. The moment she pasted something in and did not. None of that is captured by a document turned in on Sunday night. All of it is what a teacher would actually need, in September, to know what her students are doing while the software runs in the other tab.

The training gap is real. It will get closed, slowly, by August in-services and vendor certifications and confident memos from the front office. The evidence gap will not close on its own. A teacher can be perfectly trained on the tool and still walk into October with no way to see what happened between the assignment and the submission. That is the shape of the room the ninety-to-eight ratio is about to open into.

What The Fall Will Actually Ask

Ninety percent of students are already fluent in a tool the adults around them have not yet been given the instruments to read. The urgent question this fall is not whether more teachers can be trained faster. It is whether they will be given something to look at when they walk into class on Monday morning.

The instrument that shifts the ratio is not another certificate. It is the record of the hour itself, made legible to the teacher whose confidence, this July, is sitting at 2.53 out of five.²

If nine out of ten of your students are using AI and you cannot yet see what they are doing with it, what is your first week of school actually measuring?