AI in Education Needs a Relational Infrastructure

‍TLDR: Genuine AI literacy must be grounded in meaningful social practice.

‍From the classroom to the living room, and from Silicon Valley to the halls of Congress, the national conversation about AI in education has gone from a din to a clamor. This conversation is largely instrumental in its focus — which tools to use, how to prepare teachers, how to detect misuse, and how to update policy. But the research on how people learn, adapt, and form judgments suggests we are overlooking a critical dimension: the relational and social one.

AI literacy: Offloading versus thinking

When AI literacy is taught as individual tool use, students will develop habits of offloading rather than habits of thinking. Genuine AI literacy emerges through social practice — debating AI outputs with peers, questioning results with teachers, co-authoring and then critiquing together. The social context is not a nice-to-have; it's the mechanism of internalization.

Learning from our earlier ed tech mistakes

Providing AI access without building relational infrastructure risks replicating the laptop gap mistake of the early 2000s. Students in well-resourced environments don't just get access — they get AI embedded in networks of mentors and informed peers. Schools serving lower-income students must actively build that social capital, or AI adoption will widen the gaps it promises to close.

Preparing students to thrive in an AI-shaped world means preparing them to understand AI's social effects: how it reshapes trust, authorship, and community knowledge. Schools are the only institutions positioned to develop these civic competencies at scale.

As we meet the challenge of accelerating responsible AI adoption and use in schools, we must make it sustainable, equitable, and genuinely educational. That requires investing in the relational infrastructure that converts AI access into AI wisdom.