Students on AI: “Teach Us How to Use It!”

Summary for Educators

Based on Tony Frontier
Educational Leadership, September 2026, Vol. 84, No. 1, pp. 22–26

Tony Frontier went directly to students to ask a question schools too often debate without them: What actually helps students use AI responsibly? His findings, based on in-person focus groups with more than 250 juniors and seniors at 20 U.S. high schools, suggest that neither silence nor blanket prohibition is working especially well. (ASCD)

THE BIG IDEA

Students are already living in an AI-saturated world. In Frontier’s focus groups, about 85 percent reported using AI for schoolwork, and a majority estimated that they use it for roughly 25–40 percent of their coursework. They described productive uses—getting feedback, generating practice questions, summarizing notes, clarifying concepts—as well as obvious shortcuts and outright cheating. (ASCD)

Their message to educators is surprisingly straightforward: Teach us how to use it.

Students said that when teachers ignore AI, simply say “don’t use it,” or assume that AI cannot complete sophisticated academic work, students often conclude that adults do not understand the technology. By contrast, students respond positively when teachers explicitly demonstrate what AI can and cannot do, establish realistic boundaries, explain the reasoning behind those boundaries, and show students how AI can support rather than replace learning. (ASCD)

The implication for school leaders is significant. AI literacy should increasingly be viewed not merely as a technology issue, but as part of instruction, academic integrity, assessment design, and student learning.

KEY TAKEAWAYS

• Students are already using AI extensively. Schools cannot build policy around the assumption that students will simply avoid it. (ASCD)

• Silence creates its own policy. Students told Frontier that when teachers say nothing about AI, many interpret that silence as permission to use it however they wish. (ASCD)

• Blanket bans leave too many gray areas. Grammar tools, autocomplete, summaries, embedded chatbots, and AI search functions make “no AI” increasingly difficult to define or enforce consistently. (ASCD)

• Students want teachers to demonstrate responsible AI use. Effective examples included doing work independently first and then asking AI for feedback, critiquing AI-generated errors and bias, and teaching students how to ask questions that deepen understanding rather than merely produce answers. (ASCD)

• Busywork invites shortcuts. Students said they were especially likely to turn to AI when assignments seemed repetitive, irrelevant, or disconnected from later classroom discussion. (ASCD)

• Relationships still matter enormously. Students reported being less likely to cheat in classrooms where they trusted the teacher, felt safe admitting confusion, and believed the teacher cared more about learning than points. (ASCD)

WHY IT MATTERS

The AI debate can easily become a debate about software when the deeper issue is instructional design.

Frontier’s students repeatedly connected inappropriate AI use to conditions educators already understand: excessive workload, low-value assignments, unclear expectations, an obsession with grades, weak teacher-student relationships, and assessments that reward answers without requiring students to demonstrate their thinking.

That means the most effective response to AI misuse may not be better surveillance.

It may be better teaching.

Students specifically valued classrooms where teachers required them to show their work, justify answers, draft portions of essays in class, discuss their reasoning, and explain what they had learned. These practices make student thinking visible—and make outsourcing the entire intellectual process to AI far less useful. (ASCD)

There is also an important distinction between using AI to avoid thinking and using AI to improve thinking. Schools that teach students that distinction are likely to be better positioned than those that rely entirely on prohibition.

LEADERSHIP ACTIONS

1. Replace “AI allowed/not allowed” with clearer instructional categories.
Teachers might specify, assignment by assignment: AI prohibited, AI permitted for feedback, AI permitted for brainstorming, or AI encouraged with disclosure.

2. Make AI literacy part of the curriculum.
Students should learn how AI can hallucinate, reproduce bias, oversimplify information, flatter users, misrepresent sources, and provide confident but incorrect answers.

3. Ask every department to identify productive AI uses.
Responsible AI use in mathematics may look different from its use in English, science, social studies, art, or career and technical education.

4. Examine assignments that invite outsourcing.
Ask faculties a challenging question: If a student can complete this entire assignment by pasting the prompt into AI, what exactly are we assessing?

5. Reduce dependence on AI detectors.
Frontier reports that students worry about inaccurate detection, inconsistent enforcement, and false accusations. Designing assessments that make student thinking visible may provide stronger evidence than attempting to determine which sentences “sound like AI.” (ASCD)

6. Teach a simple learning-first AI routine.
One teacher in Frontier’s study used an especially useful sequence:

Do your own work → check with AI → evaluate its response → explain what you learned. (ASCD)

LEADER REFLECTION

Frontier ends with perhaps the most important student voice in the article. One senior observed that AI had been “literally everywhere” throughout his high-school experience, yet students had received remarkably little instruction about using it responsibly. His request was not for unrestricted access. It was for adults to provide guidance about what AI should be used for, how it should be used, and when it should not be used. (ASCD)

For school leaders, that may be the central question:

Are we spending more energy trying to keep students away from AI—or teaching them the judgment they will need when no teacher is standing beside them?

Original Article

“Students on AI: ‘Teach Us How to Use It!’” by Tony Frontier in Educational Leadership, September 2026 (Vol. 84, #1, p. 22-26)

------------------------------

Prepared with the assistance of AI software OpenAI. (2026). ChatGPT (5.2) [Large language model]. https://chat.openai.com