Prompting Is Not Pedagogy

In the rush to make AI usable, many professional learning sessions begin, and sometimes end, with prompt engineering. Educators are shown frameworks, templates, structures for getting “better results.” It feels actionable. Concrete. Immediately transferable.

Don’t get me wrong. It is helpful.

But prompt fluency is not the same thing as instructional fluency.

You can learn quickly how to extract stronger outputs from AI, but that does not ensure the learning experience those outputs support is well designed.

Prompting Without Pedagogy

Better prompts tend to produce clearer explanations, more polished materials, and faster drafts. They increase efficiency and improve surface quality.

What they do not determine is the cognitive architecture of the lesson itself.

Prompting does not establish the depth of thinking required.
It does not define what kind of reasoning students must practice.
It does not determine whether a challenge is meaningful or merely procedural.

Prompting amplifies the instructional design that precedes it.

If the underlying task is shallow, more sophisticated prompts will simply generate shallow work more efficiently. If the learning goal lacks clarity, AI will dutifully produce content around that ambiguity.

The tool does not correct weak design. It accelerates it.

As we’ve explored previously: AI exposes the strength, or fragility, of the learning experiences we create.

Adding more context will improve results. It sharpens the output. But sharpening a vague goal still does not make it meaningful. If the design lacks clarity, AI will faithfully refine that lack of clarity. Context is not curriculum.

Curriculum Design Must Lead

Before educators write a prompt, they must be clear about what they are designing.

What kind of thinking should this lesson cultivate?
How should it build on what they learned last class? And how should it lead them to what they’ll need to grasp for the next?
How can the material be presented so it genuinely captures student curiosity?
Where is productive struggle necessary?
What decisions should students be required to defend?
What evidence would actually demonstrate understanding?

These questions must precede and supersede tool selection.

When pedagogy leads, AI becomes a support within a deliberate design. When tools lead, pedagogy quietly adapts itself to what the tool makes easy.

This distinction ultimately determines whether AI will deepen learning or merely accelerate production.

Human-Centered Teaching in an AI-Rich World

Teaching is not a technical optimization problem. It is relational and developmental.

Students are not inputs to be processed. They are learners navigating identity, confidence, motivation, and growth. Professional judgment includes knowing when to scaffold, when to challenge, when to slow thinking down, and when to introduce additional support.

AI can assist with planning, differentiation, and rehearsal. It can help teachers generate examples or worksheets, anticipate misconceptions, and explore multiple explanations of a concept.

It cannot, however, replace attunement. It cannot sense when a student needs encouragement rather than acceleration. It cannot determine when friction is productive or when it becomes discouraging. AI cannot read the room.

Human-centered teaching remains the anchor.

Teacher AI Literacy Must Come First

Much of the anxiety around AI focuses on students. But before debating student use, educators must develop clarity about AI’s role in their own professional workflow.

Used thoughtfully, AI can strengthen teacher practice: iterating on lesson structure, exploring alternative explanations, refining scaffolds, differentiating lessons or materials to support all learners, or stress-testing an assignment.

In these contexts, AI sharpens professional thinking rather than substituting for it. The goal is not to outsource educator expertise, but to extend it.

Since it’s all too common to perceive a sagely presence on the other side of chatbot requests, I find myself consistently reminding teachers that when using AI, they are the leaders, and AI must follow them.

It’s a line that bears repeating. Frequently.

A Higher Bar for AI Literacy

If AI literacy is reduced to prompt engineering, we’ve set the bar too low.

True AI literacy for educators includes conceptual understanding of learning, ethical awareness, recognition of in-built biases, critical evaluation of outputs, and the professional judgment to align tools with instructional intent.

Prompting is a technical skill within that broader framework. Custom bots and advanced features are tools that can support that framework.

They are not the framework itself.

Workshops that focus exclusively on prompt techniques, even with ethical considerations, risk producing technical fluency without pedagogical depth.

If AI integration is to strengthen learning rather than simply accelerate production and efficiency, professional development must build learning design capacity and judgment alongside tool skills.

Students are already experimenting with AI. Many will acquire functional fluency quickly. Access and experimentation move faster than policy or professional learning.

But technical fluency is not the same as wise use.

What students need is not a teacher who can generate slightly better prompts – they need a teacher who can model when to trust, when to question, and when to step away from the tool altogether.

Professional authority in an AI-rich classroom will not be secured by speed. It will be secured by judgment.

A Reframe Worth Holding

Good prompts do not equal good teaching.

AI fluency is not measured by syntax alone.

In an AI-rich world, the defining professional skill is not writing better prompts. It is knowing when, why, and whether to use them at all. And knowing how to push back when AI begins to derail educator expertise.

The lesson from prior waves of technology is clear: tools do not transform learning. Teachers do.

Pedagogy must drive tool use.

That has always been true.

It simply matters more now.