Ever since artificial intelligence first started catching educators’ attention, there have been numerous questions about its use - from how it could cut workload to concerns around plagiarism by students.

These issues remain relevant, but as our understanding of the technology grows, we also need to start asking deeper questions - not least concerning how we ensure that students understand it, and know how and when to use it.

It’s something we at Globeducate are building into thinking for the year ahead, with the creation of an “AI Graduate Profile”.

The aim is to ensure that all students, regardless of age, critically engage with AI - not simply as a series of checklist activities to complete, or platforms they can use, but in deeper, more nuanced ways that will serve them well whatever they go on to do.

To achieve this, we divide students into three categories:

1. Up to age 11: Curious and Safe

Students who are starting to engage with AI outputs, questioning what they see and developing the habits of mind to think about AI responsibly.

This might start with the question, “Should a computer be allowed to decide who gets a prize or a place on a team?”

Pupils talk about this in pairs before sharing with the class. They are encouraged to understand that there is no “right” answer; it is about discussion.

2. From 11 to 14: Critical and Informed

Students who interrogate AI outputs rather than accepting them, and who identify bias, notice omission and understand how these systems are built and by whom.

This might involve students interrogating AI-generated content on a topic they are studying, then being asked what it has missed, where it is wrong and whose perspective is absent.

For example, an AI-generated text on the relationship between social media use and mental health in teenagers might mention that this is an issue primarily affecting girls. Students would be asked to consider what assumptions might be built into this claim, and what might be missing from this framing.

They could also be asked to identify a group of young people for whom social media might be genuinely beneficial, and to explain why their experience is absent from this argument.

3. From 14 to 18: Agentic and Ethical

Young people who direct AI with clear intent and purpose, who choose when to use it and when not to, and who take full responsibility for those choices.

This is akin to using AI as a Socratic tutor - a thinking partner that challenges and questions rather than merely delivering answers.

That progression - from curious, to critical, to agentic - is not arbitrary; it reflects what cognitive science tells us about how thinking develops, and what our own teachers have observed about where students actually are.

Of course, this will rely on teachers being able to guide students through these stages, and we have ensured that every school can begin, regardless of where its teachers currently are, and that the journey has a clear direction for those who are ready to go further.

A four-score scale

A network of AI Champions across our schools engages with teachers and provides support and ideas as required.

To track progress, students will be graded “emerging”, “developing” or “secure” against four key areas:

• Questioning: does the student interrogate AI outputs or accept them without question?
• Purposeful use: does the student make deliberate, reasoned decisions about when and how to use AI?
• Ethical awareness: does the student show awareness of fairness, bias or privacy issues in AI systems?
• Communicating reasoning: can the student explain why they used AI in a particular way, or why they chose not to?

It is the first time we have been able to point to something beyond confidence surveys.

While we have not yet run a formal trial of the profile itself, the structured programme of classroom-based research we are launching across our schools from September is specifically designed to test and refine it in real classrooms.

This can help us to move AI pedagogical practice from a “cottage industry” in isolated classrooms to something evidence-led and shared. We will continue to refine the profile as our understanding grows.

At its heart, the programme is intended to help students move from simply asking themselves “Can I use AI for this?” to “Should I, and why?”

Richard Human is head of AI at Globeducate, a network of international schools across 11 countries