Rereading your notes feels like studying. Highlighting feels like studying. The research says both are among the weakest things you can do with an hour, and that two other habits do most of the work: testing yourself and spacing that testing out over days.
AI did not change that finding. It changed the setup cost. Making a deck of flashcards from a 50-page chapter used to take an evening. Now it takes about 10 seconds, which means the evening can go to the part that works: recalling.
What the research actually says
Practice testing beats rereading, and the gap grows with time. In a 2006 experiment, Roediger and Karpicke had students read short passages and then either reread them or take a recall test. On a test five minutes later, rereaders did slightly better. One week later, the students who had tested themselves remembered far more. Rereaders had also been more confident. Feeling ready and being ready are different signals.
Two techniques stand out across hundreds of studies. Dunlosky and colleagues reviewed ten common study techniques in 2013 and rated only two as high utility: practice testing and distributed practice. Rereading, highlighting, and summarization were rated low. A 2021 meta-analysis by Donoghue and Hattie, covering 242 studies, put the same two techniques on top.
The best gap between sessions depends on how far away the exam is. Cepeda and colleagues analyzed 317 experiments on spacing in 2006 and found that the interval that produces the best retention gets longer as the time until the test gets longer. In a 2008 follow-up, the same group found that gaps of roughly 10 to 20 percent of the time until the test worked well. Exam in 10 days? Review every day or two. Exam in 10 weeks? Weekly is fine.
One caveat belongs here. Most of these studies measured factual recall. For deep problem solving, testing yourself still helps, but it is not the whole job.
What AI changes, and what it does not
AI changes how fast you can build the practice. Upload the PDF, paste the YouTube link, or record the lecture. Notes, flashcards, quizzes, and a mind map come out the other side.
AI does not change who has to do the recalling. A 2025 study of roughly a thousand high-school students makes the point sharply. Students who used an unrestricted AI assistant during practice did better on practice problems and then scored 17 percent lower on the exam than students who studied without it. A version of the assistant that explained without giving answers did not hurt exam scores.
The lesson: let AI build the deck and explain your misses. Do not let it answer the cards for you.
A spaced schedule you can actually keep
Fixed days are easier to keep than an algorithm you have to trust. This one is built for an exam about two weeks out. Stretch or compress it to match yours.
| Day | What to do | Time |
|---|---|---|
| 0 | Generate the pack. Read the Smart Notes once. Fix anything wrong. | 20 min |
| 1 | Flashcards. Sort every card into Got it or Review. | 15 min |
| 3 | Quiz. Ask why each miss was wrong. | 20 min |
| 7 | Flashcards, Review pile only. Redraw the mind map from memory. | 20 min |
| 10 | Quiz again. Flashcards on anything still in Review. | 15 min |
| 13 | Visual Note and one short flashcard pass. Then sleep. | 10 min |
Notice what is missing: a reread day. The only reading is day 0.
How to run it in Lernix AI
Day 0: build the pack. Add the material in Lernix AI. Pick the study language before you generate if you want the pack in a different language from the lecture. Read the Smart Notes once. If the notes miss something the professor said, edit them. Cards are only as good as the notes behind them.
Flashcard days: recall before you flip. Say the answer, then tap to reveal. Swipe right for Got it only if you produced the answer. Swipe left for Review if you only recognized it. The mastery rate at the end is a map of where the next session goes, not a score to feel good about.
Quiz days: miss, then ask. The multiple-choice and true-or-false quizzes give instant feedback. For each miss, open Ask AI on that material and ask why the correct answer is correct. It answers from the material you uploaded, so the explanation stays tied to your course. Do not ask it for answers to graded work.
Mind map day: the blank-page test. Open the mind map, study its shape, then close the app and draw it from memory. Reopen and look for the missing branch. That branch goes back into the Review pile.
Last day: cues, not content. The Visual Note is a one-page infographic of the 3 to 6 core points. Save it to your photos and scan it. It is a cue, not a new source.
Make the cards better before you drill them
- One fact per card. If a card asks two things, split it in the note and regenerate, or just answer both parts out loud.
- Watch for wording cards. A card you can answer because you recognize the phrasing is testing memory of the sentence, not the idea. Rephrase the question in your head before answering.
- Check numbers against the source. Generated notes are a draft. Dates, formulas, and names deserve a look at the original PDF before they are drilled into memory.
- Keep materials separate. One material per lecture or chapter. Then the flashcard results tell you which lecture is weak instead of averaging everything together.
Four mistakes that undo the method
- Massing everything into the night before. Spacing is the technique. One long session is the thing it beats.
- Marking Got it after one correct flip. A card earns Got it when you produce the answer cold on a later day.
- Reading the notes when the plan says flashcards. If the app is open and you are not answering something, you are rereading.
- Letting the chat answer the cards. Explanations after a miss help. Answers before an attempt remove the recall, and recall is the point.
Frequently Asked Questions
Does Lernix AI schedule my reviews automatically?
No. Lernix does not send review reminders or compute due dates for you. Flashcard sessions sort cards into Got it and Review, and you can rerun any session whenever you like. Use the day schedule above, or put it in your calendar.
How many flashcards should I study in a session?
Enough to get through the Review pile honestly. If you are flipping without recalling, stop and come back. Short, honest sessions on separate days beat one long one.
Are AI-generated flashcards accurate?
They are built from the material you upload, and they are a draft. Check them against the original, especially for numbers and definitions, and edit the notes if something is wrong.
Flashcards or quizzes: which is better?
Both are practice testing. Flashcards are faster for definitions and facts. Quizzes with instant feedback are better for spotting misconceptions, because the wrong options are designed to be tempting. Use both on different days.
Where can I read the research?
Dunlosky, Rawson, Marsh, Nathan, and Willingham, Improving Students’ Learning With Effective Learning Techniques, Psychological Science in the Public Interest, 2013. Roediger and Karpicke, Test-Enhanced Learning, Psychological Science, 2006. Cepeda, Pashler, Vul, Wixted, and Rohrer, Distributed Practice in Verbal Recall Tasks, Psychological Bulletin, 2006. Cepeda, Vul, Rohrer, Wixted, and Pashler, Spacing Effects in Learning, Psychological Science, 2008. Donoghue and Hattie, A Meta-Analysis of Ten Learning Techniques, Frontiers in Education, 2021. Bastani and colleagues, Generative AI Without Guardrails Can Harm Learning, PNAS, 2025.
Start with one lecture
Open app.lernixai.com, or get Lernix AI on the App Store and Google Play. Upload one chapter, read the notes once, and put flashcards on the calendar for tomorrow.
For specific inputs, see PDF to flashcards and YouTube to study materials. For a full exam week, see how to study for midterms with AI.