How to Build a Study Plan with AI

A study plan answers three questions: what to study, when, and how. AI handles the first two well if you give it real inputs, and it can nudge you toward better answers to the third. Start with the facts only you know: - the exam date - the syllabus topics and how much each is weighted - how confident you feel about each topic, on a 1 to 5 scale - the hours you can realistically study each day - fixed commitments such as work shifts Research helps with the third question, how to study. In a 2006 study, Henry Roediger and Jeffrey Karpicke found that students who practised recalling a passage remembered more of it a week later than students who reread it, even though the rereaders expected to remember more. A 2013 review led by John Dunlosky compared ten techniques. It rated practice testing and distributed practice (spreading study over time) as the most useful, and rated rereading and highlighting low. So a good AI plan should schedule self-testing rather than just 'review chapter 4', and return to each topic several times with growing gaps. A sensible plan has three phases: 1. A first pass through all the material. 2. Mixed retrieval practice, with more time on weak topics. 3. One or more full timed practice exams, then lighter review in the last day or two. Build in a spare day each week so one missed day does not throw off everything after it. Watch for three misconceptions. A detailed plan is not the same as a realistic one. A model that has never seen your course cannot know which topics your instructor emphasises. And calendar arithmetic is a weak spot for AI: language models often miscount days or give a date the wrong weekday, so compare the plan with a real calendar.