The best use cases for AI in schools depend on human judgment before the tool ever gets involved.

A district leader can ask AI to draft a lesson plan. But the tool still needs direction about the curriculum, the standard, the student work, the level of cognitive demand, the scaffolds that are appropriate, and the evidence of learning that matters.

A principal can ask AI to summarize walkthrough notes. But the tool still needs a clear instructional lens. Otherwise, it may organize observations neatly without helping anyone understand whether students were doing the right kind of thinking.

A curriculum leader can ask AI to draft assessment items. But the tool still needs a definition of alignment, rigor, and acceptable evidence.

That is the point. AI is not a substitute for instructional leadership. It is a pressure test for it.

If the district's instructional vision is clear, AI can help leaders and teachers work faster inside that vision.

If the vision is vague, AI can produce faster versions of the same inconsistency.

AI can support a lot of work, but it should not be asked to make the decisions leaders have not made.

It cannot decide:

Those are leadership decisions.

They require a point of view about learning, curriculum, assessment, teacher support, and student opportunity. They also require enough discipline to keep the system focused when a new tool, program, or initiative enters the conversation.

UNESCO's guidance on generative AI in education makes a similar point from a policy angle: schools need a human-centered approach, public engagement, safeguards, and capacity building. That matters because AI adoption is not only a technical decision. It is an instructional and organizational decision.

For district leaders, the practical question is not, "Should we use AI?"

The better question is, "What instructional work are we trying to strengthen, and what boundaries do we need so AI supports that work instead of distracting from it?"

One of the tricky things about AI is that weak outputs can look impressive.

A lesson can be well formatted and still be misaligned to the standard.

A student task can sound engaging and still avoid the actual content students need to master.

A feedback comment can be encouraging and still fail to tell a teacher what to do next.

A district memo can sound clear and still avoid the hard instructional tradeoffs underneath the initiative.

That is why districts need more than access to AI tools. They need a way to judge AI outputs against a shared instructional vision.

This is where many districts should slow down before they speed up.

Not because AI is unimportant. It is important.

But if leaders scale tools before they clarify expectations, they may create a system where everyone is using AI differently, producing different materials, and reinforcing different definitions of quality.

The result is not innovation. It is inconsistency with better formatting.

Before a district moves from experimentation to broader AI use, leaders should answer four questions.

What is our instructional point of view?
Name what strong instruction should look like in the district. This does not need to become a 40-page framework. It should be clear enough that leaders, coaches, and teachers can use it to make decisions.

What should AI help us do better?
Start with a real instructional or leadership need. For example: improving checks for understanding, strengthening lesson internalization, drafting better family communication, summarizing walkthrough evidence, or helping leaders prepare for coaching conversations.

Where does professional judgment need to lead?
Decide which tasks require a human decision before, during, or after AI support. This is especially important for curriculum alignment, student data, personnel decisions, student support, and feedback to teachers.

How will we review quality?
Create a simple process for checking AI outputs against district expectations. The review should focus on instructional quality, not just whether the output is fast, clean, or convenient.

These questions help leaders avoid two weak positions: blocking AI completely or adopting it without instructional guardrails.

The stronger position is this: use AI where it can reduce friction, but keep instructional judgment in the hands of educators and leaders who know what students need.

Here is one useful exercise for a district leadership team.

Pick one common task. For example:

Then ask AI to complete the task.

Do not stop with the first output. Review it against the district's instructional expectations.

Ask:

That last question is the most important one.

When an AI output misses the mark, the problem may be the tool. It may also be the prompt. But sometimes the prompt is weak because the district's instructional direction is not clear enough yet.

That is not a reason to panic. It is useful information.

AI can become a mirror. It can show leaders where the system has shared language, where expectations are clear, and where the district is still relying on assumptions.

Do not start with, "How can we use AI?"

Start with, "What instructional work are we trying to make stronger?"

Then decide whether AI can help.

That sequence matters. When AI comes after instructional vision, it can support better planning, communication, analysis, and follow-through. When AI comes before instructional vision, it can create a lot of activity that looks productive but does not improve the core work.

AI will not replace instructional vision.

But it may expose where instructional vision is missing.

If your district is trying to sort through where AI belongs in the work, start there. Clarify the instructional direction first. Then build the tools, prompts, routines, and guardrails that help people act on it.

UNESCO: Guidance for generative AI in education and research
A useful policy-oriented starting point for human-centered AI use, safeguards, and capacity building in education.

Stanford HAI: 2024 AI Index Report
A broader look at AI trends, capabilities, public perception, and responsible AI issues.

Stanford SCALE repository: Generative AI and Its Educational Implications
A helpful education-facing overview of AI applications, limitations, and the need for educators to understand both.

If your district is experimenting with AI, start by asking where the tool is creating clarity and where it is revealing confusion.

If that question is live for your team, contact us at [email protected] to schedule an initial consultation.