Three AI curricula agree on ethics but start above age ten

The classroom note

A ninth-grader building a web app ‘with AI as a creative partner’ and a ten-year-old sorting pictures to train a simple model are both doing ‘AI literacy,’ but the two lessons share almost nothing except the label. Three organisations publish curricula openly enough to compare: Code.org's AI curriculum, now rebranded CodeAI, MIT RAISE's Day of AI, and Common Sense Education's AI Literacy Lessons. Reading what each actually teaches, and to which grade, shows both real convergence and a real gap.

What the evidence says

Code.org's own curriculum pages set out grade-banded courses: ‘How AI Makes Decisions’ for grades 3-5 uses hands-on predictive-data-model activities; ‘Artificial Intelligence Discoveries’ for grades 6-8 centres on understanding, directing and creating AI ethically; ‘Artificial Intelligence Foundations,’ now a full-year pathway for grades 9-12, moves from how AI works to building web apps with an AI teaching assistant. MIT RAISE's Day of AI reports reach rather than content depth: more than two million learners across 175 countries and 70,000 trained teachers, which establishes scale, not what any lesson covers. Common Sense Education's collection is the most granular: discrete 15-to-20-minute lessons pinned to grades, from ‘Remix Responsibly’ at grade 5 through ‘Artificial Intelligence: Is It Plagiarism?’ at grades 9-12, with named lessons on AI bias, chatbot design, facial recognition and deepfake consent. None of the three publishes evidence that completing its lessons changes a student's later judgement about AI; each states its own content and reach, not a measured outcome.

The implementation question

Laid side by side, the youngest grade any of the three curricula names is grade 3 (Code.org) or grade 5 (Common Sense Education); none publishes a lesson for children younger than that. The three also divide the ethical ground differently: Code.org folds ethics into its coding units, ‘Our AI Code of Ethics,’ ‘Societal Impact of Generative AI’; Common Sense Education isolates it into short, discussion-based lessons on chatbot friendship, deepfakes and plagiarism; MIT RAISE foregrounds global reach and teacher training over a single ethics unit. A school choosing between them is choosing an emphasis, not a neutral menu of the same content.

What holds and what fails

What holds is that all three, independently, treat how the technology works and what it means to use it responsibly as separate strands that both need teaching, a point of agreement across organisations with different funding and audiences. What fails is any claim that AI literacy is now comprehensively covered across ages: the published examples cluster from grade 5 upward, leaving early years and lower-elementary students, already meeting AI products directly on this batch's other evidence, outside all three curricula's stated scope.

• Which grade does your school's chosen curriculum actually start at?
• Does it teach how the tool works, how to use it responsibly, or both?
• What happens for younger children who are already using AI products the curriculum does not yet address?

Three well-documented curricula agree that AI literacy needs teaching from around age ten; none of them extends that agreement down to the ages where AI products are already arriving unaccompanied by a lesson plan.