Assessment, when the work can be generated.

For a long time a submitted assignment was its own evidence: if it was in front of you, someone had done the thinking. That is no longer true. The work still arrives on time and reads well. It just doesn’t tell you who did the thinking.

So Noeta asks the one question that still works: can this student explain what they turned in?

The common answer to generated work is a detector. It guesses whether AI was involved, and it guesses wrong often. The students it accuses wrongly are usually the ones with the least standing to argue, and on most campuses that means multilingual students.

That is the failure that makes a tool unusable on a campus: confidently telling a professor a student cheated when they didn’t. Once it has happened to one student, every report after it is read as the next accusation.

So Noeta reports what it heard, with the student’s own words behind every finding and the recording one click away. Whether any of it is a misconduct matter is a judgment only you can make.

A student can paste a prompt into a chatbot and read the reply back. What is far harder is holding a live, unscripted conversation about their own specific choices: why this answer, in these words, and not the other one.

So Noeta speaks each question aloud rather than showing it, and asks for the reasoning behind an answer. Near the end she poses a case that was not on the quiz, built on the spot from what the student just said, so there is nothing to look up. A workaround is still possible. It is slower, and it leaves more behind, than the shortcut it replaces.