A position paper and a prompt guide answer different questions

The classroom note

A teacher opening a lesson-assistant chatbot for the first time does not need a position statement; they need to know what to type. That gap between policy and practice shows in two public-sector documents published for English schools. The Department for Education's position paper on generative AI, first published in March 2023, sets out considerations for schools at a general level. Oak National Academy, a DfE-funded curriculum body, instead published a guide of specific example prompts for its own AI lesson assistant, Aila, aimed at the exact question a position paper does not answer.

What the evidence says

The DfE document establishes scope rather than instruction: it names the technology it covers, generative AI tools and large language models such as ChatGPT and Microsoft Copilot, and says it is aimed chiefly at compulsory-age education, pointing readers to a companion document on product safety for technical detail. It does not itself supply prompts. Oak's guide does exactly that: it walks a teacher through ten stages of building a lesson with Aila, from the learning outcome through misconceptions to an exit quiz, and lists suggested prompts verbatim, such as asking Aila to ‘break my explanation down into smaller chunks’ or make a task ‘require less reading.’ Both are living guidance rather than evaluated interventions; neither measures whether following them improves a lesson.

The implementation question

What Oak's guide assumes, and states outright, matters as much as what it recommends: Aila is built with its own helper, a content moderation agent that reviews the assistant's output before a teacher sees it, and the guide warns that some appropriate content may still need sensitive handling, flagged rather than removed. That is an admission, from the tool's own publisher, that a specific prompting guide does not remove the need for a teacher's own judgement afterward; it structures the request, not the review. A single teacher's quoted estimate of saving around 30 minutes per lesson appears in Oak's own materials as a testimonial, not a measured result, and should be read as one person's account rather than a general finding.

What holds and what fails

What holds is the division of labour the two documents unintentionally demonstrate: a department can set boundary conditions, which tools, which safety expectations, while a curriculum body closer to daily practice supplies the prompts teachers actually use, and neither substitutes for the other. What fails is treating either as sufficient alone: a position paper without example prompts leaves a teacher guessing at wording, and a prompt guide without an independent evaluation of the lessons it produces leaves a school unable to say whether the saved time changed what students learned.

• Does your school follow a department position paper, a specific prompting guide, or neither?
• Does the tool your school uses review its own output before a teacher sees it, the way Aila's guide describes?
• Has anyone measured what a prompt-built lesson produces, beyond the time it took to build?

A position paper tells a school what to consider; a prompting guide tells a teacher what to type. Neither tells anyone what the finished lesson is worth.