Classroom surveys point in a consistent direction: teachers use generative tools mostly for preparation, not instruction. Lesson plans, differentiated worksheets and feedback drafts dominate.
Student use is harder to measure and easier to moralise about. Schools that redesigned assessment around process rather than product report fewer disputes than those that banned the tools outright.
The equity question is unresolved. Paid tools are markedly better than free ones, which reproduces existing advantage unless institutions buy licences centrally.
What has not happened is replacement. The scarce resource in education remains attention from an adult who knows the child.
