AI is a productivity tool. Manage it through policy: integrity, data privacy, vendor review, acceptable use.
This is where most district AI policy lives today.
AI is an environment. Each new medium restructures how children remember, attend, reason, and know.
This is the framing K-12 has not yet adopted — and the one this talk argues we need.
Ong gave us oral, literate, and broadcast. The current moment requires two further stages — one for algorithmic distribution, one for algorithmic generation.
A K-12 classroom in 2026 is operating across all five at once.
Ong's framework rested on three things we can no longer take for granted.
The condition in which algorithmic systems both curate and generate the symbolic content children learn inside — making human authorship optional at scale.
Kapur (2016) gives us a 2×2 that names the precise risk of generative AI in K-12.
Correct output. No schema construction. The artifact looks like learning — the learning never occurred.
Correct output and genuine understanding. The student wrestled with the material.
Wrong output, no growth. Task exceeded the learner's reach without scaffolding.
Wrong output, real learning. Struggle activates prior knowledge and prepares future learning.
The artifact of learning remains visible. The process of learning becomes optional.
Locates the problem in a few students who cheat. Solves it with detection tools, honor codes, locked browsers.
Necessary, but insufficient — and aimed at the wrong target.
Locates the problem in the media environment itself, whose default tendency is the bypass of cognitive labor.
Students aren't lazy. The path of least resistance now leads to unproductive success.
The intentional preservation of the cognitive, dialogic, and existential resistance that durable learning requires — when frictionless automation has become the default condition of the medium.
Three phenomenological, one structural. Together they specify what K-12 has to protect.
The internal struggle of comprehension, synthesis, and revising one's own thinking. Where schemas actually get built.
The dialogic struggle of defending claims to unpredictable human interlocutors. Discussion, peer review, oral defense.
The vulnerability of being held personally accountable for one's own thinking, in physical space, in front of others.
District policy, assessment design, prof. dev., the values communicated through institutional practice.
"Preserve struggle" without this distinction is a recipe for reproducing inequity under the language of rigor.
A multilingual student uses an AI translator to convert a draft from their first language into English. What just happened?
The student can demonstrate content knowledge and join academic discourse without being filtered through a language barrier unrelated to the learning objective.
Translation can produce grammatically correct, stylistically generic prose — removing the perspective and phrasing that make a child's writing recognizably their own.
No mechanical rule resolves this. The same tool reduces both kinds of friction at once. Distinguishing them requires situated, professional judgment — which is what the framework asks K-12 educators to develop.
Friction has to become a structural value, not a personal one — or it will not survive.