AI tutoring expands in schools, but human teachers remain essential

AI-powered tutoring platforms are rapidly entering classrooms, promising highly personalized instruction that adapts to each learner's pace. Research, however, shows that human tutors still achieve stronger learning gains and that AI works best when it supplements, not replaces, skilled teachers. Effective use therefore requires policies that keep educators at the centre of instruction while leveraging AI for feedback, differentiated practice and expanded learning opportunities.

Scaling educational AI also presents technical challenges. Unlike consumer recommendation systems, errors in learning platforms can take weeks to surface, affecting credential timelines and confidence. A recent incident at a large online programming course caused a temporary infrastructure overload, preventing thousands of learners from accessing required workspaces and disrupting a key learning moment. The episode underscores that reliable compute infrastructure is part of the instructional experience and must be designed with learning science and robust feedback loops in mind.