With truly personalized learning I would expect more than incremental returns on the traditional pedagogical approach. I learn very rapidly with LLMs at my fingertips, largely due to the dialectical approach -- that is, having a conversation with the model allows me to dive in to places that are blurry and uncertain. Rather than "read, digest, quiz" maybe with the new tooling there is a better way to zero-in on what is unclear other than doing an assessment. Just kinda thinking out loud here.