Early learners are disappearing into AI’s cognitive Divide
GUEST COLUMN | by Chris Holoka
I recently spoke with a high school English veteran of fifteen years from a small rural district in central Michigan about her frustrating experiences with AI. What really keeps her up at night, however, isn’t in the classroom. It’s at home, with her second-grade daughter, now surrounded by an AI ecosystem that threatens to “rescue” her from the struggle that creates strong readers.
As AI takes hold in schools, the digital divide, once a gap in access, usage, and digital literacy, has quietly expanded to include something less visible but more consequential: a cognitive scaffolding divide.
‘As AI takes hold in schools, the digital divide has quietly expanded to include something less visible but more consequential: a cognitive scaffolding divide.’
The shift is redefining educational equity for the AI era. This gap isn’t measured in devices or bandwidth but in the interaction itself, in how the system guides, constrains, or replaces a learner’s thinking. Good scaffolding builds toward metacognition, the learner’s awareness of their own thinking, and none of it develops without the productive struggle AI is increasingly engineered to eliminate. Equity in AI isn’t about access to answers. It’s about access to the struggle required to understand them.
Where the Divide Hits Hardest
The divide runs deepest where thinking itself is still being built. For early learners, the internal machinery for self-monitoring, error detection, and metacognitive awareness is not yet running. Without it, there is no backup system.
That inner scaffolding isn’t the only thing missing. So is the way in, as early learners are functionally disconnected from prompt-based AI. The keyboard is a closed door. They cannot type a query, refine a response, or direct an exchange. Today’s dominant AI platforms cut a single effortless path: user prompts, system responds. That path needs to be inverted: system leads, learner follows. AI won’t be a tool they use; it will be their learning environment. The difference is agency. Older students have it, pre-readers don’t.
The early literacy window is narrow. Pre-K through second grade is when developing minds are primed for exactly the kind of thinking AI is built to skip. The Science of Reading is unambiguous: phonemic awareness, decoding, and fluency must be constructed, not received. They require effort. They require failure. They require trying again. By default, AI is an escalator past all of it. By answering before the learner has struggled, supplying before they have retrieved, completing before they have decoded, the model cheerfully smooths over any resistance that may build an actual reader.
The best outcomes belong to learners taking the stairs. They’ll encounter decoding, retrieving, and deciding on the way up, and that encounter is the point.
A Blueprint for Development
Early learners need a fundamentally different AI architecture than the too-familiar prompt-response design. A content vending machine, built on asking and receiving, assumes a user who can formulate a request. A pre-reader cannot, which means the entire interaction must be rebuilt. Here’s where to start:
Proactivity
A system that doesn’t wait for input that isn’t coming. It leads. The learner follows, and in following, begins to build the capacity to lead. It should meet them right at the edge of what they’re almost able to do and hold that space until confidence makes it unnecessary.
Multimodality
The text channel is effectively closed. Well-designed tools communicate through visuals, audio, or even haptics, not as enrichment layered on top of a text interface, but as the interface itself.
Intentionality
Each interaction targets something specific, a discrete reading skill like phonics or fluency, or a metacognitive strategy like self-monitoring or making predictions. A gifted tutor doesn’t address everything a child hasn’t learned; they diagnose, isolate, and work one skill at a time.
The system initiates, the interface delivers, the interaction targets. Each product decision at every level is an instructional decision, shaping what is practiced, skipped, and never encountered. The design is the curriculum.
Who Gets to Think?
Product teams: build something that wouldn’t panic a fifteen-year veteran teacher who knows enough about AI to lie awake worrying about her second grader. Something that doesn’t require a keyboard. If your earliest learners can’t participate with a single tapping, dragging, and probably sticky little finger, you haven’t built it for them.
Educators and administrators: the feature list isn’t the right scorecard. Before adopting an AI tool, ask what thinking it will replace.
The first wave of edtech equity was about access. The next wave will be about cognition. Technology is shifting from passive resource to cognitive participant. The question is no longer “what can the AI do?” but “who is being asked to think?” The danger isn’t that students rely on AI. It’s that they never develop the ability to know when they shouldn’t.
—
Chris Holoka is an edtech executive and product leader with 20 years of experience at the intersection of AI, design, and learning outcomes. He served as VP of Product Design and Product Management at Learning A-Z, building literacy tools for millions of K-12 learners. Reach him at chris@holoka.com.
0 Comments