From AI guidelines to classroom decisions

What does the work need to show?

A rule about an AI tool is only useful if it connects to the learning a task is meant to develop or assess. The same kind of assistance can have different consequences in different subjects, and at different stages of learning. A generic instruction to disclose AI use leaves an important question unanswered: does the submitted work still demonstrate the student’s own understanding?

This project developed subject-specific student AI-use and academic-integrity guidance, then brought it into a teacher workshop. My role included designing and coordinating the framework and its movement into practice. Subject teachers contributed expertise, review and validation. The subject knowledge and professional judgements involved were a collaborative contribution.

The framework described here is school-specific practice. It is not presented as official IB policy, a universal ban on AI or a claim that every school should make identical decisions.

Protect the reasoning being assessed

In Theory of Knowledge, the core evidence is the student’s reasoning about knowledge: making and justifying claims, considering alternatives, explaining examples and reflecting on their own thinking. A sophisticated-looking paragraph is not sufficient if the student cannot explain how its ideas were reached.

The TOK guidance therefore distinguishes learning support from replacing the thinking being assessed. During learning, assistance may sometimes help a student question an idea they have already developed. That possibility depends on prior independent effort, the task’s purpose and the student retaining ownership of the reasoning. Having a tool supply the line of argument before that effort changes what the activity is asking the student to do.

Disclosure remains necessary where AI has been used, but it cannot by itself restore missing evidence. Acknowledging that a tool produced the evaluation does not demonstrate that the student can evaluate. The framework treats honesty about assistance and validity of assessment evidence as related questions that both need an answer.

Make the boundary understandable

Professional documentation needs to become language that a student can use during a real task. The Individuals & Societies version groups guidance under three headings: “Usually OK”, “Ask your teacher first” and “Not OK”. Task-specific instructions still take precedence; the categories are a starting point for judgement.

A source-evaluation task makes the distinction clear. Asking for an explanation of what a source’s purpose means may support access to the task. Asking AI to evaluate the sources’ value and limitations supplies the very judgement the student is supposed to practise. Both requests concern the same topic, yet they do different work for the learner.

Other uses sit between those examples. Summarising a source or proposing a structure might help in one activity and remove essential learning in another. The middle category gives that uncertainty somewhere to go: a conversation with the teacher before proceeding. Clear boundaries need an equally clear way to ask a question.

Look at the development of the work

Authentication becomes more informative when it includes the process through which a piece of work develops. The TOK guidance identifies draft development, planning, in-class writing, oral explanation and teacher–student dialogue as possible sources of evidence.

The relevant question is whether a student can explain and reconstruct substantive parts of the work. An oral discussion can ask why an example was selected, what supports a claim or where an argument has limits. These approaches require professional judgement; none is presented here as an automatic verdict about authorship.

The design implication is to consider evidence of learning when setting a task, not only after receiving its final product. Expectations about independent work, permitted assistance and opportunities to explain the process belong together. Their implementation must also remain understandable and fair to students.

A workshop built around decisions

The orientation workshop moved from documents to practice. Subject groups reviewed student-facing wording against the fuller guidance, looking for ambiguity or missing distinctions. They then discussed a subject scenario and considered what they would do and why.

The purpose was to make room for professional disagreement and explanation. It was not a technology demonstration or a search for one answer detached from subject context. Teachers were invited to share something the discussion had raised, reinforced or might change in their practice.

Teacher use of AI for planning is a separate question from student use within assessment. The purposes and responsibilities differ. In both cases, the relevant judgement begins with the work being done, the evidence needed and who remains responsible for checking the result.

A precise account of progress

As of 14 September 2026, the English Section orientation workshop had completed the main creation, consultation and validation process. Full professional, student-facing and parent-facing guidelines were complete. Posters and remaining communication and display steps formed part of final implementation. Communication and implementation with the Georgian Section were still outstanding at that date.

That supports a claim of a developed framework and delivered workshop, with rollout continuing. It does not establish full schoolwide implementation, consistent application or a measurable reduction in misconduct. The next substantive question is how the guidance informs teaching and assessment decisions over time. No impact evaluation is claimed in this account.