Microsoft Research has launched StudentSim, a research system aimed at judging whether tutoring content actually helps without large-scale human trials. The system uses public de-identified learner records to train a personalized AI student simulator for each student, modeling both independent answering and how students respond after receiving tutor explanations. The original text says it addresses the limits of AI tutors that focus only on answer correctness and of costly, hard-to-scale real-learner studies, supporting faster model iteration.
According to Wired, Maka, a US streaming app for children aged 0-6, has launched, featuring content selected by child development experts. Every episode is evaluated through a framework co-developed with researchers at the Yale Child Study Center, and the official site says the library currently holds more than 500 episodes. Education research organization SRI has licensed its APPROVE AI video analysis technology to Maka to identify early math and literacy content, with AI screening first and experts reviewing each episode afterward. Maka is currently available only in the US, at $14.99 per month or $107 per year, with a 14-day free trial.
