Every few weeks another institution announces a crackdown on AI use: detectors, declarations, threats of disciplinary action. I understand the instinct. But I think we are fighting the wrong battle.
AI detectors are unreliable, and they tend to punish the students who can least afford to be wrongly accused, including those writing in a second or third language. A tool that guesses is a poor basis for a decision that can affect someone's academic record.
More importantly, if a chatbot can complete your assignment in thirty seconds, the real question is what that assignment was measuring in the first place. A take-home essay that asks for a summary of a well-known topic was already easy to outsource, to a classmate, a past paper or a paid writer. AI just made the weakness visible to everyone.
The alternative is not complicated, but it takes effort. Ask students to apply ideas to a local case they have actually seen. Ask for drafts and reflections on how their thinking changed. Add short oral defences, even five minutes, where they explain their own work. Grade the process as well as the product.
None of this is perfect, and none of it scales effortlessly to a lecture hall of four hundred. But assessment that rewards understanding is the only kind that survives every new tool we have not met yet.
Policing treats AI as a threat to be contained. Redesign treats it as a fact of life and asks what we truly want students to be able to do. I know which conversation I would rather be having.