Teacher judgment is the centre of the response.
Generative AI can draft, summarize, translate, reorganize, and imitate. Those capabilities are useful, but they do not determine what is worth learning, what students should practise independently, or what counts as credible evidence of understanding.
Those remain professional decisions. The central question is therefore not simply, “How can we use AI?” It is, “What kind of learning are we responsible for protecting and producing?”
Pedagogy first. AI second.
Assessment must make thinking visible.
When polished written products can be generated rapidly, schools need richer evidence: conversations, oral explanation, observation, drafts, reflection, source use, revision, and performance over time. This does not mean abandoning writing. It means designing assessment so that authorship and reasoning are visible.
Curriculum coherence matters more, not less.
AI makes it easy to create activities. Ease of production can produce fragmented curriculum unless teachers begin with standards, conceptual goals, sequencing, and a clear account of what students should be able to do.
Ethics belongs inside instructional design.
Privacy, bias, accuracy, authorship, intellectual property, access, and environmental cost should be considered when a task is designed, not appended as a warning after the fact.