Giving educators an AI account is not professional learning. It is the beginning of a workplace change.
San Francisco Unified has deployed ChatGPT Edu to 12,000 teachers and staff while still developing its districtwide AI policy. Its public guidance says it is not board-approved and may no longer be current; a final policy vote is scheduled for March 2027. Recent reporting describes teachers and parents asking for more evidence, transparency, and participation as the district expands AI-enabled tools.
The lesson extends well beyond San Francisco. Districts often treat access, training, policy, and evaluation as separate workstreams. Educators experience them as one change. If the license arrives before the purpose is clear, if training explains buttons rather than judgment, or if evidence is collected only after scaling, people will fill the gaps with either enthusiasm or suspicion. Neither is a strategy.
Professional learning around AI must be embedded in real decisions: Which task is worth changing? What should remain human? What new risk or workload appears? What evidence would justify continuing?
Before your next AI expansion, convene a compensated, cross-role design cohort—teachers, leaders, IT, special education, and labor—to test a few high-value uses for one cycle. Publish the use cases, boundaries, success measures, and what changed because of educator feedback before scaling.
In Human Still Required, I put it this way: “AI rewards efficiency. Leadership requires intention.”
The question is not how quickly staff can get access. It is whether the system is learning as deliberately as the people it expects to change.
Has your district funded licenses—or funded learning?