National survey results from the last school year show that most teachers who used AI received no formal training, and roughly half taught themselves through their own exploration. Only about one in five worked at a school with an established AI policy. The result is a patchwork in which skilled and careful users sit alongside teachers who are unsure what is allowed. Schools that want responsible use need a way to bring everyone to a shared baseline.

Self-teaching has strengths. Teachers who experiment often discover creative and efficient uses, and they build confidence through practice. It also has weaknesses, including uneven understanding of privacy, accuracy, and fairness issues, and a lack of shared language for discussing what works. A structured program can keep the benefits of experimentation while addressing the gaps, and a program that treats existing teacher experiments as valuable raw material tends to earn more cooperation than one that starts from scratch.

A ninety-day roadmap is long enough to build habits and short enough to maintain momentum. It divides the work into three phases that move from understanding to practice to shared standards. Each phase has a small number of concrete activities and outputs. The goal is not expertise in every tool but dependable judgment, and many schools start the clock at the beginning of a semester so that the final phase lines up with the planning for the next term.

Days 1 to 30: Build a shared foundation

Begin with a short session that covers the school's policy, privacy basics, and the main risks of inaccuracy and bias. Share a one-page summary of what is allowed and what is not, and give teachers a way to ask questions. Ask each teacher to try one low-stakes task, such as drafting a rubric or a differentiated reading, and bring the result to a brief share-out. This phase builds confidence and surfaces questions.

• Start with a one-page summary of school policy and privacy basics
• Assign one low-stakes experiment in the first month
• Review anonymized essays with draft feedback in small groups during month two
• Create a short checklist for reviewing AI-assisted feedback
• Finish with shared standards and a plan for ongoing peer coaching

Days 31 to 60: Practice with real student work

In the second month, move to tasks involving student writing. Pair teachers in small groups to review anonymized essays alongside draft feedback from an approved tool, scoring them against a common rubric first. Ask them to identify which comments were accurate and useful and which needed correction. This hands-on work builds the habit of review and develops a shared sense of what quality feedback looks like.

Introduce a simple checklist for reviewing AI-assisted feedback, covering accuracy of scores, relevance of comments to the student's actual writing, tone, and privacy. Have teachers apply it to their own work for one assignment and report on what they found. Collect common problems and discuss solutions. Practical experience converts abstract advice into skill, and the checklist can be revised as teachers discover which items catch the most mistakes in real feedback samples.

Days 61 to 90: Set shared standards and plan next steps

The final phase turns experience into shared agreements. Hold a department meeting to decide how much review is required for different types of assignments, what to tell students and families, and how to handle errors. Draft or revise your school's guidance based on what teachers learned. Make sure the resulting document is short, clear, and specific, and publishing the result as a one-page guide on the staff site makes it easy for substitutes and new hires to find.

Plan ongoing support, such as a monthly thirty-minute meeting where teachers share examples and questions. Identify a few teachers who can serve as peer coaches. Decide how you will measure whether the training worked, perhaps through a brief survey and a review of feedback samples. Learning should not stop at day ninety, and a short anonymous survey at the end of the term will tell you whether teachers feel more confident and where they still need help.

Include everyone, including skeptics

A sizable minority of teachers are uneasy about AI, and some oppose it in the classroom. Their concerns about independent thinking, fairness, and privacy are legitimate and valuable. Invite them into the conversation, ask them to help write the checklist, and respect their choices about use where policy allows. Programs that listen to skeptics tend to produce wiser guidance, and including a skeptic on the planning committee often leads to a checklist that catches problems enthusiasts would miss.

Make participation feel professional and not punitive. Offer time during the school day, recognize contributions, and avoid framing the program as a remedy for deficiency. Teachers who feel respected will engage more deeply. A shared sense of purpose is the most powerful driver of change, and a brief note of thanks from leadership after each session shows teachers that their time and effort are noticed, while keeping the tone collaborative instead of corrective.