The Leftover Question: Recovering Student Help-Seeking When AI Answers First

Before generative AI tools, a student who got stuck had little choice but to ask someone. Those questions were never plentiful, but they told instructors much: which step was shaky, which term misread. That information was never designed into a course. It was a byproduct of difficulty, produced whenever a student got stuck enough to ask. Generative AI tools remove the getting stuck, and the byproduct goes with it. Nothing visibly breaks. Submitted work looks better, fewer students appear stuck, and instructors lose their sense of where the class is. This Quick Fix describes The Leftover Question, a routine designed to recover part of that loss for three minutes of class time. After any task where help was available, students write the question they still have, naming where the tool’s answer felt vague or beside the point. They post it to a shared, low-stakes space before class, anonymously if they prefer. The next session opens by reading two or three aloud and answering them. Students are never asked what they used or typed. What is collected is a question, not a record. The instructor begins the session knowing something specific, instead of guessing from work a tool may have improved.