AI Is Moving Into Classrooms — This Paper Says It Needs Rules

AI has quietly become a teaching assistant in classrooms. It grades code, answers student questions on forums, generates practice diagrams, and coaches writing style. But most schools adopted these tools without a clear answer to a basic question: when the AI gets something wrong, who is responsible? A new paper by researcher Ethan Dickey argues that the fix isn't a policy memo written after the fact — it's designing the rules into the tool itself.

His idea, called "instructional governance by design," says a teaching tool's limits should live in how it works, not in a handbook nobody reads. He proposes six questions to grade any classroom AI: Does it follow real teaching methods? How much authority does it have over a student's grade or learning? Can a human override it? Does it push students to think, or just hand over answers? Is it built for this specific course? And can anyone see how it reached its conclusions?

He then tries the framework on real tools used in computing and first-year engineering courses. A tool that simulates a teaching assistant uses grading rubrics. A code helper nudges students to reflect instead of copying answers. A forum-answering system routes replies through staff review before posting. A diagram generator works under a human teaching assistant's supervision. Each one needs different guardrails, because each does a different job with different stakes.

For parents, students, and anyone hiring graduates, the takeaway is practical. AI grading and tutoring is spreading fast, and "the AI said so" is not an acceptable answer when a grade, a transcript, or a skill is on the line. The paper offers schools and companies a shared vocabulary for deciding how much rope to give these tools — and where a human must stay in the loop.