My washing machine says “AI” on it.
So does my dryer. The box told me the AI would detect fabric type, load size, and soil level, and adjust the cycle accordingly. My car has AI-assisted lane keeping. My phone's keyboard is AI-powered. Somebody is going to sell me an AI-enabled toaster within the next eighteen months and I am going to buy it, because I am tired.
I have been teaching high school engineering for 22 years. When I started, finding an answer was expensive. You went to the library. You asked the teacher. You skimmed the index of a textbook. Answers lived in specific places, and getting to them cost time.
Twenty-two years later, answers cost almost nothing. Most of what my students learn in a high school classroom can be retrieved from ChatGPT in six seconds. Gemini will explain the derivative of x-cubed more patiently than I can. An AI can write a lab report that looks convincingly like a student wrote it. My washing machine can tell the difference between jeans and a delicate blouse.
In a world where answers are cheap, questions hold the most value.
And I think the teachers who figure that out first are the ones whose students are going to need them the most.
Answers Used to Be the Product
For most of the history of formal education, a teacher's primary job was delivery. Transfer information from one brain to many brains, then verify the transfer with a test. The whole apparatus of school — the rows of desks, the bell schedule, the worksheets, the Scantron sheets — was optimized for that one function.
That function is now essentially free.
I am not being dramatic. Any kid with a phone has access to a better explanation of quadratic equations than most of the human beings who have ever taught quadratic equations. The information transfer problem is solved. It took about eighteen months.
A lot of teachers look at this and panic. Some of us look at it and think: finally.
What We Actually Teach
Here is a thing I have been trying to say out loud for a few years now. The best teachers I have ever worked with, the ones whose students come back twenty years later to thank them, did not really teach information. They taught a relationship to information.
They taught:
- how to notice when an answer does not actually address the question you asked
- how to tell the difference between “I understand this” and “I can repeat this”
- how to ask a follow-up question that moves a conversation forward
- how to sit with not knowing something long enough to find a good question about it
None of that is threatened by a chatbot. All of it is harder than it sounds, takes years to develop, and is exactly the kind of thing a kid cannot learn from a prompt box.
The content was never really the content. The content was the scaffold we used to teach kids how to think. The scaffold still matters, so I am not telling anyone to throw out their curriculum. But the curriculum was never the point. The habits of mind we taught around the curriculum were the point.
Questions Are the New Curriculum
If we want to stay useful to students in an era when every fact in the world is one keystroke away, we teach students to ask better questions. Here is what that actually looks like in a classroom.
Asking why before how
An engineering student wants to build a bridge for a class project. They ask ChatGPT, “what is the strongest bridge shape?” ChatGPT says triangular truss bridges are generally considered the strongest for their weight. The student writes that down. Done.
But the real question is “strongest for what load, over what span, made of what material, under what failure mode?” The AI will happily answer that too, if you ask it. Most students will not ask it, because they do not know that is the interesting question. That is what we teach.
Questioning the AI's answer
A student turns in an essay. I suspect ChatGPT wrote most of it. Instead of playing detective, I ask, “okay, explain to me why the third paragraph is wrong.” They stare at me. They do not know what is wrong with it, because they did not write it, and the AI does not tell you what its weak spots are. You have to ask.
The next week I teach them how to find the weak spots on purpose. “Ask the AI a follow-up question that exposes an assumption you do not agree with.” That is a skill. It is not in any state standard I have ever seen. I think it should be.
Rubrics that reward the right question
I used to grade lab reports on whether the student got the right answer. Now I grade them on whether the student asked a question the data could actually answer. Did you set up an experiment that would tell you something you did not already know? Or did you go hunting for a predetermined conclusion?
The first kind of student is going to be fine no matter what AI does next. The second kind is going to get replaced by a browser extension.
The Knowledge Gap Is Really a Question Gap
I built ClassLens partly because I wanted my grading time back, and partly because I wanted to see something I could never see before: the shape of what my whole class did not understand, all at once.
When ClassLens grades a class set, it generates a Knowledge Gap Report. The usual way teachers think about that is “here are the things my students got wrong.” But that is not really what it is. What it is, is a map of the questions my students did not think to ask.
They wrote down a number without units because they never asked “what are the units for this?” They skipped the error analysis because they never asked “how do I know if my answer is right?” They answered the surface question on the rubric because they never asked “what is this rubric actually testing?”
Those are all question failures. You cannot fix a question failure by showing a student the right answer. You can only fix it by teaching them to notice the question they skipped in the first place.
That is what grading is for now. Not to measure whether a student can repeat an answer an AI could have given them faster. To surface which questions they did not know were there.
Try One of These This Week
If any of this is landing, here are three small things to try before Friday.
- Give students an AI-generated answer and ask them to find what is missing. Not what is wrong. What is missing. It is a harder question and a better one.
- Rewrite one line of one rubric.Change it from “correctly identifies X” to “asks a productive question about X.” See what happens.
- End a lesson with “what do you still want to know?” instead of “any questions?” The second one closes a door. The first one opens one.
We Were Always the Question People
None of this is new. Socrates was doing it. The best teachers you ever had were doing it. What is new is that it is no longer optional.
The kids who come out of our classrooms knowing how to ask good questions are going to be fine in a world full of cheap answers. The kids who come out knowing only the answers are going to be competing with my washing machine.
Teachers were never the answer people. We were always the question people. The world is finally catching up.