A transferable teaching pattern · ECON 2316 example

Argue With the AI

The AI defends a claim that sounds right but hides a flaw. The student's job: use course theory to catch it. Learning happens in the evaluation, not the answer.
The pattern (any discipline): give students an AI that confidently states a defensible-but-wrong position → they must rebut it with course concepts → they grade the AI's reasoning. Here it's micro theory; swap the claim for your field.
AI
Perfect price discrimination eliminates deadweight loss — output reaches the competitive level. So it's efficient, which means it's the best outcome for society.
Student
Careful — you conflated efficiency with welfare. Yes, perfect price discrimination removes the deadweight loss, so it's allocatively efficient. But it does that by transferring all consumer surplus to the monopolist — consumers are left with zero. "Efficient" only means the surplus isn't wasted; it says nothing about who gets it or whether that's good. A distributional judgment is separate from efficiency.
AI
You're right. I overstated it — efficiency (no DWL) is not the same as desirability. The equity question is distinct, and I collapsed the two.

AI-THINK

Student evaluates and corrects AI output using theory — the core AI-readiness skill.

AI-DATA

Sees a concrete limit of LLMs: fluent, confident, and still wrong.

AI-ETHICS

Efficiency vs. equity — models can smuggle value judgments into "neutral" claims.

AI-CREATE

The human adds the judgment the AI lacked — the value-add that stays theirs.

Adapt it: History — AI defends a tidy but contested narrative; students cite sources. Philosophy — AI argues an ethics position; students find the hidden premise. Policy — AI recommends a "efficient" policy; students surface who bears the cost. Same pattern, your discipline.