Adaptive curriculum

AI can take feedback from humans, including answers to questions and work created by the student, to modify future lessons.
This would likely take quick computation as future lessons would need to be modified on the fly as the student is learning.

Excerpt from a ChatGPT Essay on this topic:
The next frontier of AI in education lies in expertise and personalization. Instead of being confined to general-purpose chat interfaces, advanced AI could act as a subject-matter specialist, as competent as a university professor in mathematics, physics, history, or literature. But unlike human experts, this AI would not be constrained by time, fatigue, or class size. Every learner could have access to a private mentor of the highest caliber. Imagine a student struggling with calculus being guided step-by-step through concepts, exercises, and examples—each tailored to their preferred learning style, pace, and even emotional state. The AI could adjust explanations dynamically: offering visual diagrams to one student, real-world analogies to another, and rigorous proofs to a third.

Beyond one-on-one explanations, AI could build entire personalized curricula. Unlike static textbooks or one-size-fits-all syllabi, these curricula would continuously adapt, measuring progress, identifying gaps, and adjusting difficulty in real time. Drawing from insights in cognitive science, the AI could optimize learning schedules, spacing practice and review for maximum retention. The result would be a system that not only teaches but actively engineers growth, using data from each learner to refine its teaching methods. This represents a shift from passive instruction to active optimization of human learning.