Two things happened this year that most parents and many principals have not caught up with. In April, CBSE published a Computational Thinking and AI framework for Classes 3 to 8 and advised schools to start delivering it from the 2026-27 session. In July, Punjab became the first state to make AI a mainstream subject from Class 1 to Class 12. AI literacy has stopped being an after-school club and started being a line on the timetable.
That is good news, and it raises a harder question. What should a school actually teach, and how would anyone know whether a learner learned it?
In 2024, UNESCO published an AI competency framework for students. It is the first international reference for what school-age AI education should cover, and it is more thoughtful than most of the courses being sold to schools right now.
It organises AI competency into four aspects: a human-centred mindset, the ethics of AI, AI techniques and applications, and AI system design. Each runs through three levels, understand, apply and create. That gives twelve competencies. The first two levels are for every learner. The third is for those who want to build.
Three of its instructions matter more than the rest.
First, it deliberately does not break the content down by grade. It tells curriculum developers to build a spiral: the same four aspects revisited every year at greater depth, so a learner meets ethics, data and design at eight, again at eleven, and again at fifteen.
Second, it warns that a course delivered as a one-off, a hackathon or a bootcamp, may spark interest but is unlikely to produce lasting competency. Spaced, repeated engagement is the design requirement.
Third, it asks that AI education be product-agnostic. A learner should understand how a system learns from examples, not how to click through one vendor's tool.
One more thing worth knowing. UNESCO does not certify curricula. Any course described as "UNESCO certified" is describing its own claim. The honest form is a published mapping: this objective serves that competency.
If you are a principal or a teacher being offered an AI programme, these are the questions that separate a curriculum from a bundle of activities.
Does it spiral across years, or is it a set of workshops? A single term of robotics kits is an event. A curriculum returns to the same ideas at increasing depth.
Is ethics woven through, or bolted on at the end? Fairness, privacy and accountability belong inside the lesson on how a model learns, because that is where bias comes from.
Is it independent of any product? If the lessons only work inside one company's app, the learner has been trained on a tool, not educated about a field.
How is the create level handled? Building and defending a real system is project work with a rubric and a teacher. A multiple-choice quiz cannot assess it, and a curriculum that pretends otherwise is measuring the wrong thing.
And the question that decides everything: what evidence of learning will you have at the end of the year? A certificate of attendance says a learner was present. Evidence says what they can now do, competency by competency, on dated work.
Alea now has an AI literacy track that runs from Class 3 to Class 12 as a single spiral, mapped competency by competency to the UNESCO framework. We will publish the mapping, so the claim can be checked rather than taken on trust.
Classes 3 to 5 sit at the understand level and are taught in stories and sorting games: what a helper that learned from examples is, why it can be unfair if the examples left people out, and why your information is yours. Classes 6 to 8 follow the twelve Early AI modules CBSE has specified, which is the part of the CBSE framework most schools have no content for beyond the free handbooks. Classes 9 to 12 move into apply and create: first Python on data the learners collect themselves, hands-on classification and regression, a community project, and a capstone defended to a panel that includes someone from the community it serves.
Every objective is taught by Alea's tutor one to one, with a mastery check at the end that is dated and kept. Projects are planned with the tutor and assessed by the teacher. Because every objective carries its UNESCO competency, a learner's progress can be read competency by competency rather than as a percentage on a certificate. Nothing in the track names a vendor. Where learners use a tool, it is a free one the teacher chooses, and Alea assesses what they understood from it.
A teacher sets up an AI literacy class in a few minutes, adds learners by name, and hands out PINs. Each learner's first sign-in is a placement check that puts them at the right class, which is sometimes not the one they are in.
The framework is explicit that AI should support teachers, not replace them. We agree, and the design shows it. The create level is assessed by a person, not a quiz. The teacher sees each learner's placement and progress from work actually done, rather than from a record filled in from memory. And the class stays the teacher's class.
If your school is working out how to deliver AI literacy this year, we would like to talk. Write to us at info@alealearning.com.
Start with a free placement and feel the difference a tutor that adapts to you actually makes.