A few months into my Cohort 2 fellowship with Kutunga.org's A-Codes programme, I recently sat in a session led by John Zoltner, founder of AIChildSafety.org. It was one of those sessions that quietly rearranges how you think about a problem. Zoltner has spent years at the intersection of child protection and technology, and the way he frames AI has stuck with me since [Zoltner, AIChildSafety.org].
I left that session with a simpler question than the one I walked in with. Not how do we keep AI away from our children, because that ship has sailed. The better question is how we equip them to use it well.
Not whether, but how
For a long time, the instinct in education has been to treat AI as a threat to be managed. Block it, ban it, restrict it in exam season. Zoltner's point cuts through that instinct. Children are already meeting AI, in the search bar, the homework helper, the voice assistant on a parent's phone. The real work is not keeping AI away from them. It is building the scaffolding that helps them make better choices inside that relationship, at every age.
Cognitive scaffolding is not a new idea in teaching. It means giving a learner just enough support to do something they cannot yet do alone, then pulling that support away as the skill takes root. Apply that to AI, and the job of a teacher changes. An educator becomes less of a gatekeeper and more of a guide, someone who helps a child ask better questions of the tool in front of them.
What that looks like, stage by stage
In the early years, scaffolding looks like conversation. Researchers describe AI as a kind of "black box" to young learners, an unfamiliar system they meet without understanding how it works [ETS Research, Preparing K-12 Students With AI Literacy, 2025]. Why did it suggest that? Could it be wrong? These questions build a critical instinct long before a child ever types a prompt.
In the middle years, scaffolding tightens around ownership. Students can use AI directly now, but research on AI-supported writing warns that leaning too hard on automated feedback erodes a student's sense of authorship, along with the productive struggle real learning depends on [AI-Integrated Scaffolding to Enhance Agency and Creativity in K-12 English Language Learners, MDPI, 2025]. A student who never wrestles with a blank page never learns to fill one.
By high school, the scaffolding should be coming down. Older students have the reasoning skills to interrogate AI output rather than accept it, treating it as something to build with and argue with, if the years before gave them the practice [Lin, Dai & Ng, Constructionism in K-12 AI Literacy Education, SAGE, 2025].
Whose framework, whose children
This is where my fellowship experience matters most to me. Kutunga's WATOTO Framework, built from the documented voices of 599 children across 14 African countries, insists that no product or policy for children is complete without community-level accountability, not just individual protection [Kaberi & Makumbe, The WATOTO Framework, Kutunga Design Academy & Innovation Lab, 2026]. That principle belongs in classrooms too. A teacher scaffolding AI literacy is not acting alone. They are standing in for a whole community's responsibility to a child, which is a very Ubuntu way of putting it.
We cannot avoid AI. But we can decide, deliberately and at every age, how much of the thinking we let it do for our children, and how much we insist they keep doing themselves. That is the real work of AI literacy, and it belongs to educators as much as to technologists.
Comments(0)