Back to Bigger Picture

BIGGER PICTURE Learning & work

Twenty-five minutes back from the science planning pile

A useful teaching assistant does not have to design an entire lesson. In an English school trial, teachers mainly asked ChatGPT for a quiz, a fresh activity or a clearer explanation—and saved preparation time.

AI-generated illustration: A teacher reviewing an organised science lesson plan.
A teacher reviewing an organised science lesson plan. Conceptual illustration; not a photograph or an exact depiction of the real project.AI-generated illustration of the idea, not a photograph of the actual people, equipment or event.
Trial
259 teachers in 68 schools
Weekly preparation
56.2 vs 81.5 minutes
Measured scope
Year 7–8 science lessons

How the idea works

  1. Select one lesson ingredient

    The teacher chooses a quiz, explanation or activity.

  2. Generate and adapt

    AI proposes material; the teacher checks and revises it.

  3. Track actual preparation

    Weekly diaries compare time for the defined lessons.

One useful ingredient in a lesson

Teachers used the tool selectively: generating questions, adapting materials for their class or finding an activity idea. Most did not ask it to plan everything. NFER’s account describes support for one or two elements of a lesson. This is a relatable use of AI: relieve a repetitive preparation task while leaving the teaching decisions with the teacher.

Sources: [2]

Practice came before the measurement

The randomized trial involved 259 science teachers in 68 English schools. The AI group received a practical guide; the comparison group prepared without generative AI. Five initial weeks allowed teachers to learn the approach. The main time comparison then covered weeks six to ten, rather than assuming benefits on the first day.

Sources: [1][3]

A modest change, a measurable saving

For the Year 7 and 8 science preparation measured, weekly time averaged 56.2 minutes with ChatGPT and 81.5 without: 25.3 minutes, or 31%, less. Teachers generally used the saved time for other teaching work or lower workload. The figure concerns this part of their timetable, not a 31% reduction in their whole working week.

Sources: [1]

The teacher remains the quality filter

An expert panel, unaware of how the sampled resources were made, found no evidence of a quality difference. That limited sample does not prove equivalence or better pupil learning. The practical lesson is to preserve checking and adaptation: a faster first draft becomes valuable only when the teacher can confidently use it with a particular class.

Sources: [2][3]

The original sources

  1. Education Endowment FoundationTeachers using ChatGPT can cut lesson planning time by over 30 per cent
  2. National Foundation for Educational ResearchNFER Classroom: ChatGPT for lesson planning, research summary
  3. National Foundation for Educational ResearchWritten evidence to the Education Committee, AIE0146

What to explore next

A maths tutor that fits inside a WhatsApp conversationA clinical-trial search that explains each eligibility rule
Explore all 99 applications