The IRIS Scale asks schools to be clear about how AI is used
Fundació Escola Cristiana de Catalunya has introduced a six-level guide that makes the roles of people and artificial intelligence visible in schoolwork.
On July 21, 2026, Fundació Escola Cristiana de Catalunya (FECC) introduced the IRIS Scale, a guide for the use of artificial intelligence in school tasks. Its central idea is straightforward: before debating whether AI is allowed or banned, schools should be able to explain what the tool did and what the person did.
The proposal is not a public rule for every school in Catalonia, nor is it a cheating detector. It was prepared by FECC's digital-expertise group, coordinated by Miquel Àngel Prats, a professor of Educational Technology at Blanquerna–Ramon Llull University. It is meant to give schools, teachers, students, families, leadership teams, and administrative staff a shared language for pedagogical decisions.
Six ways to work with AI
IRIS places tasks on six levels, from 0 to 5. Level 0 means no AI is used. At level 1, AI acts as a corrector: it may check spelling, translation, or mechanical wording without contributing ideas. Level 2 is the AI as an inspirer, used to suggest options or help someone begin.
At level 3, AI produces a first draft that the user must check and verify. Level 4 describes co-creation, where a person and system engage in a more sustained dialogue to develop a task. At level 5, AI produces content, always under human supervision.
The scale does not say that a higher number is better. Its own framework stresses that level 0 or 1 may be the most appropriate choice depending on the learning goal. If a teacher wants to assess whether a student understands a concept, working without AI may be essential. If the goal is to learn how to spot errors or compare phrasing, limited assistance can be appropriate.
Transparency, not surveillance
The important innovation is less the numbering than the way it is meant to be used. IRIS does not rank students or measure merit. It describes the degree of cognitive delegation between a person and AI in a particular task. The question becomes: which part was done by the student, teacher, or tool—and why?
That shift can make academic integrity more useful than a blanket statement saying “I used AI.” A research assignment may require human source selection, interpretation, and conclusions while allowing language correction. In another activity, evaluating an AI-generated draft may be the lesson itself. What matters is that the expectation is explicit before work begins.
The framework includes examples for secondary-school and upper-secondary students, teachers designing an exam, families supporting homework, leadership teams analysing data, and staff managing enrolment. It avoids treating AI in schools as one single problem.
Part of a longer effort
FECC had already published a responsible-AI protocol in March 2025 and audience-specific recommendations that July. IRIS adds a more specific focus: academic integrity in each task. The foundation plans to share the material with its member schools through training and to present it across the network at its Quòrum26 conference in October.
No scale can by itself resolve model bias, privacy risks, or the need for teacher training. But it can help schools avoid two common mistakes: treating every AI-assisted task as fraud, or mistaking any automated output for learning. Making that difference visible is a useful place to start.
Sources
This article was produced with artificial intelligence under human editorial oversight.