What is the short answer?
Start with a specific task, review the tool and the data it needs, assign a responsible person and agree on how to evaluate the result. A small pilot gives the team a chance to spot problems before extending use to more classrooms.
Which educational problem are we trying to solve?
'We want to use AI' does not define a need. 'We want to draft two versions of a reading activity and review them before class' gives the team something testable. Record the current process, the intended improvement and the decisions that remain the teacher's responsibility.
UNESCO's guidance proposes a human-centered, age-appropriate approach that considers privacy when evaluating educational AI uses. It is general guidance; a particular choice still needs review in the school's own context.
Source: UNESCO — Guidance for generative AI in education and research (2023)
What should be clear before a trial begins?
If the handling of an uploaded file is unclear, redesign the task around a fictional example while the question is resolved. Removing a student's name does not necessarily remove other identifying details.
- Purpose: one concrete task and an observable result.
- Access: who will use the tool and under which conditions.
- Data: what information is necessary and what stays out.
- Review: who checks the output before it is used.
- Incidents: who to contact and when to stop the trial.
What could a useful first exercise for teachers look like?
Take a fictional activity about the water cycle. Ask for two alternative sets of instructions. The teacher checks whether each preserves the objective, introduces errors or uses language that fits the group. Then decide what to keep, rewrite or discard.
Review is part of the work, not an extra step after completion. The draft might save preparation time, need too many corrections or add nothing useful. Each outcome helps the team decide whether that use is worth continuing.
How will we know whether the pilot helped?
Agree on criteria before starting: clarity of instructions, errors found, time spent preparing and checking, and usefulness to teachers. If students participate, the school also determines which evidence of learning is appropriate.
Asking whether people liked the tool is not enough. An entertaining activity may not improve understanding. Avoid claiming learning gains without an evaluation design that supports them. An internal pilot informs decisions; it does not establish general effectiveness.
Which training could help the team move forward?
AI with Marcé's teacher programs address three needs: exploring tools through specific tasks, designing purposeful learning activities and planning coordinated adoption. A school with a plan addresses that third conversation: priorities, boundaries, responsibilities and a pilot outline.
The workshop leaves the team with questions and decisions to take forward. It does not replace school approvals or specialist technical or legal review where needed.
A school with a plan. Explore the program and its PDF →Which sources did we consult?
How do we put these answers together?
We check factual claims against the linked sources. Activities are educational suggestions from AI with Marcé, not research findings or undocumented classroom experiences. Dates indicate publication and content changes. Email us to suggest a correction. Email Alex.