What Bounded AI Looks Like in a Writing Classroom

The Problem

Students have access to general-purpose AI regardless of school policy. The question is not whether AI reaches them (it does), but whether the AI they use in instructional contexts is connected to the teacher, visible to the school, and bounded by the assignment. Most consumer AI tools are none of those things.

Bounded vs. Unbounded AI

• Unbounded AI: general-purpose, private, no connection to the assignment or rubric. The teacher cannot see the interaction or evaluate what the student actually did.
• Bounded AI: class-linked, assignment-specific, rubric-anchored, teacher-visible. The student works within the structure the school assigned.

Instructional AI that is unbounded produces the same outcome as no instructional AI: students complete work; teachers cannot tell who learned anything.

The Governance Problem

The concern most commonly framed as academic integrity is actually an instructional visibility problem. A teacher who cannot see what the student actually produced cannot assess development, identify where instruction is needed, or build a record of what the student can do. General-purpose AI tools make that visibility structurally impossible.

What School-Governed AI Looks Like in Guided Scholar

Independent practice in school-account deployments is not private student AI use. It is a governed, class-linked practice activity visible to the responsible teacher and controlled by the school deployment context.

The teacher dashboard shows every submission: original draft, feedback delivered, revision submitted, revision depth between drafts. The system delivers rubric-aligned feedback. The teacher evaluates the student’s response and decides what happens next. That is the boundary between a tool that informs teacher judgment and one that replaces it.