According to this article, "Recent advances in AI raise the possibility of building something education has never really had: a virtual laboratory for learning. At the center of this possibility are synthetic students." The problem, of course, is that we don't know how to simulate students. "When you ask a capable model to play a struggling student, its mistakes tend to come out inconsistent or superficial."Now are we good at "predicting how a particular lesson or explanation or piece of feedback changes what a student understands and remembers later." Ultimately, I think, the article is a pitch to philanthropy: "Much of what we need, it turns out, is shared scientific infrastructure that no individual lab or company has much incentive to build today."
Today: Total: [Direct link] [Share]
