AI, curriculum and teachers
When explanation becomes cheap, teaching has to become more human and more exacting.
I use AI in education work already. The useful question is not whether a tool can generate an explanation. It is which decisions still require a strong teacher.
Where AI helps in my work
I use AI inside course and curriculum workflows to organise teacher source material, expose gaps, produce inspectable drafts, and run consistency checks before a human release decision.
That is useful work. It can make a large body of teaching easier to structure and review.
It does not turn the model into the subject expert.
The teacher remains responsible for the chemistry, the sequence, the examples, the assessment judgment, the explanation a learner will actually understand, and the final decision to publish.
The line I will not cross
AI can produce a plausible answer faster than most people can check it.
In education, plausible is not enough. A small error can be copied into a lesson, repeated in an assessment, and scaled to thousands of learners. The system therefore needs traceable sources, visible review, independent checks, and a person who owns the release.
The operating rule is simple:
Use AI to make judgment easier to apply. Do not use it to make judgment disappear.
What changes for teachers
When a recording or an AI system can repeat an explanation, the teacher’s value moves toward work that does not scale neatly:
- diagnosing the exact misconception;
- deciding whether to explain, question, practise, or pause;
- reading the student’s confidence as well as the answer;
- designing tasks that reveal thinking;
- giving feedback that changes the next attempt;
- building a serious classroom culture;
- knowing when a tool is wrong.
This is not a smaller role. It is a more demanding one.
What I can work on with a school
I am developing bounded workshops and pilots for schools and education teams around:
AI for real teacher workflows
Where AI can save time in planning, feedback preparation, resource production and review, with clear limits and human ownership.
Curriculum and course-production systems
How to turn teacher source material into a structured learning sequence without confusing syllabus coverage with learning design.
Assessment that changes teaching
How to design checks that reveal an exact problem and lead to a better next decision, instead of producing another mark that nobody uses.
Chemistry teacher development
Question-reading, misconception diagnosis, precise explanations and the relationship between subject understanding and the answer an examiner can reward.
Start with one real problem
I am not offering a generic speech about the future of AI.
A useful pilot begins with one real workflow, one group of teachers, a visible starting point, and a result the school can inspect. If the work helps, we build from evidence.
Ask about a teacher-development pilot or read When Explanation Becomes Cheap, Teaching Becomes More Human.