In a sixth-grade humanities class at a San Francisco school, teacher Tristen Chang no longer assigns essays for homework. Instead, she holds oral exams: she asks fifteen questions about characters, plot, theme and metaphors, and students answer within five minutes.
Since last year, Chang and dozens of other middle and high school educators across the country have been abandoning familiar assessment formats. They are cutting homework, moving writing and discussions into the classroom, and grades increasingly depend on how a student thinks in real time rather than on a finished text.
The mechanics are simple. A student comes to class without a phone or laptop. The teacher asks a series of questions on material studied together. Answers are recorded immediately, with no opportunity to consult a chatbot. Written work is done on paper or in class under supervision. Time for reflection is limited, so what is tested is the ability to analyze and connect ideas, not to reproduce someone else's text.
Reuters interviewed a dozen teachers and experts. More than half of students and educators already use AI for school assignments. So far these are interview and observational data, not a large randomized study. The limitation is obvious: it is unknown how such changes will affect academic performance in a year or two and whether they can be scaled in schools with different levels of resources.
The main tension is not in the technology itself, but in who gets access to genuine assessment of thinking. In classes where teachers can devote time to oral questioning and individual work, the skill of critical analysis develops. In other schools, the burden on educators grows, and the gap between those who manage to adapt and those who do not only widens.
The question now is not how to ban AI, but which forms of knowledge assessment can withstand its presence and at the same time remain fair to all students.
