AI is a multiplier, not a substitute — what that means for revision

Harvard Online's argument is simple and worth taking seriously as a student: AI raises the ceiling of what a skilled person can do, but it does very little for someone who has not built the underlying judgement. Multiplication needs something to multiply.

In revision terms, that means the value of an AI study tool depends almost entirely on what you ask it to do. Asking it to hand you an answer removes the thinking that would have formed the memory. Asking it to test you, reorganise your own notes, or explain the step you got stuck on keeps the thinking with you and removes only the admin.

Use AI for the work that isn't learning

Retyping notes, formatting a summary, splitting a chapter into sections, writing a first pass of flashcards from material you already wrote — none of that is learning. It is preparation. Automating preparation buys you more time inside the part that actually moves your grade: recall, practice and correction.

Keep the hard part manual

Retrieval has to hurt slightly to work. Answer first, then check. When you use a tutor, describe your reasoning before you ask for the correct one — the difference between your explanation and the model's is the exact thing you need to revise.

Verify against your own source

The safest AI study setup is a closed loop: your notes in, study material out, with citations back to the passage it came from. If a claim cannot be traced to your material or your textbook, treat it as a prompt to go and check rather than as a fact to memorise.

The practical version of Harvard Online's point is this: let AI compress the time between sitting down and doing real recall. Everything else it offers you is optional.