Using AI to Reduce Teacher Workload

Teachers can use generative AI to draft quizzes, suggest lesson activities, adapt a resource for another class or prepare a first version of routine communication. These tasks may become faster, but each output needs review. A tool that creates a worksheet quickly may save little time if the teacher must correct errors, rewrite dense language or repair an unsuitable activity. Time saved on one task may also be filled by other work rather than reducing a teacher’s total hours. Evidence should be read within its scope. In a randomized trial reported by the Education Endowment Foundation and National Foundation for Educational Research, 259 Year 7 and 8 science teachers across 68 schools took part in a ten-week comparison of ChatGPT-supported lesson preparation with usual preparation without generative AI. The ChatGPT group spent about 25 minutes less per week on lesson and resource preparation on average, a 31% reduction in that measured task. The study concerned lower-secondary science preparation, not every subject or all parts of a teacher’s job. Resource-quality analysis used a limited sample, and the trial did not establish that AI prevents burnout. Start with a bounded, repeated task and measure the whole workflow: prompting, checking, editing and sharing. Use approved services and avoid entering identifiable student information into an unapproved tool. Keep a non-AI method when access, policy or learning goals require it. A time log can show whether an assistant reduces workload, shifts it or adds new busywork. Teacher well-being also depends on staffing, schedules, administrative expectations and support. AI may help with a small part of preparation; it cannot solve those structural conditions by itself.