Building Inquiry Questions: Have AI Ask, Not Answer

Ask "any questions?" in class and the room goes quiet. Students can't come up with questions not because they lack curiosity but because nobody ever taught them how to build a good question. After a decade or more of training to pick the right answer, the muscle for making questions never got used. And in the age of AI, questions matter even more. AI produces answers in an instant, so a student's edge shifts to knowing what to ask. Given the same AI, the work a student produces with a narrow, sharp question and the work produced with a vague one are worlds apart. Paradoxically, you can use AI itself to build that questioning muscle, as long as you cast it as a questioning coach rather than an answerer.

How to Cast AI as a Questioning Coach

Have AI produce answers and students only copy them down. Have it produce questions and students think.

• Turn the question back: Set it up so that when a student asks something, AI doesn't answer straight away but asks back, "What happens if you narrow that question further?"
• Rank the questions: AI sorts three student-written questions into "questions a search finishes" and "questions that take thinking to answer," and shows the split.
• Practice transforming questions: AI builds examples of turning closed questions into open ones alongside the student, so the student picks up the transformation patterns.
• Chain the questions: AI nudges the student from one answer to the next question, encouraging follow-up thinking.

The better AI gets at answering, the more what classrooms must build is the muscle for asking, not for answering.

A One-Period Question-Building Lesson

Say you're covering "urbanization" in social studies. A flow with question-building at the center looks like this.

1. Question burst (5 min): Students look at the topic and pour out questions by volume. No judging good from bad. Hands don't go up at first, but set a quota, "let's write five each," and even stuck students start writing something.
2. AI sorting (5 min): Feed the poured-out questions to AI and have them sorted into closed and open questions.
3. Refining questions (10 min): Pick one closed question and, with AI coaching, transform it into an open question you can investigate.
4. Testing questions (10 min): With a partner, check whether the refined question "has something to research" and "isn't settled by a single answer."
5. Adopting a question (5 min): Choose the one question most worth investigating as the group's question.

The product of this lesson isn't an answer; it's the question itself. If the quality of student questions is better at the end than it was in the first minutes, the lesson succeeded.

Key Takeaways

Lesson design in the age of AI has to shift its weight from teaching how to get good answers to teaching how to build good questions. Cast AI as a questioning coach rather than an answerer, and students train the thinking that sorts, transforms, and chains questions themselves. Try turning the first 5 minutes of your next lesson from "find the answer" into "pour out questions." Students find it awkward at first, but once they experience their own questions getting sharper, the room starts becoming a questioning classroom.