Ace
Short answer: making your own flashcards does help you remember (the generation effect), but the bigger gains come from testing yourself on the cards repeatedly and spacing those sessions out. AI-made cards are fine if they come from your own material, you check them, and you spend the time saved answering them. The best mix: let AI draft from your notes, write your own cards for the things you keep getting wrong.
The case for making your own
There is a real finding behind the advice to write your own cards. The generation effect is the observation that people remember information they produce themselves better than information they simply read. In a meta-analysis of 86 studies and 445 effect sizes, Bertsch and colleagues (2007) put the average benefit at about 0.40 of a standard deviation, which is meaningful in learning research. The size varied a lot depending on the task.
Generating questions, specifically, has also been tested in a real university lecture. Ebersbach, Feierabend and Nazari (2020) compared students who wrote their own questions about lecture content, students who answered questions, and students who restudied. A week later, both generating questions and answering them improved recall; restudying did not.
So when you write a card, you are doing useful work: deciding what matters, putting it in your own words, and testing your understanding as you phrase the question.
What matters more than who wrote the card
Notice the second half of that study: answering questions did about as well as writing them. Retrieval is the engine. Roediger and Karpicke (2006) found that students who practised recalling a passage remembered substantially more of it a week later than students who restudied it.
Timing matters too. Kornell (2009) found that spacing flashcard practice beat cramming for 90% of participants, even though most of them believed the opposite after their first session.
This is the real trade-off. If making a deck by hand takes you three evenings, those are three evenings you could have spent answering cards across the week. A handmade deck you only review once is worse than a decent AI deck you review five times.
Where AI flashcards go wrong
- Wrong or wobbly cards. AI can state things that are not in your material or not quite right. A wrong card learned well is worse than no card.
- Off-syllabus cards. If the AI uses general knowledge instead of your notes, you get cards on things your course never covered, phrased differently from your lecturer.
- Too easy. Generated cards can end up testing recognition of a term rather than understanding. "What is osmosis?" is easier than "Why does a red blood cell burst in pure water?"
- Skipping the thinking. If you let the AI pick what matters and never look at the source, you lose the part of making cards that helps most.
Our guide to checking AI study answers covers how to spot the first two.
A hybrid that gets the best of both
- Generate from your own material only. Your slides, your notes, your set chapter. Not "the topic in general".
- Skim the deck against the source once. Delete cards that are wrong, trivial or off-syllabus. This is quick, and it is already a form of review.
- Answer before flipping. Say or write the answer from memory, every time.
- Write your own cards for your misses. When you get one wrong twice, rewrite it in your own words, or add a "why" card next to it. That is where the generation effect is worth the effort.
- Space it. Come back to the deck over several days rather than in one long session. Our spaced repetition schedule sets out the gaps.
Which should you choose?
- Make your own if the topic is new to you and you have time: writing the cards is part of understanding it.
- Use AI drafts when the exam is close, the material is large, or you already understand it and need practice volume.
- Mix them in most real situations, as above.
Whichever you choose, the source matters. Ace writes practice questions only from the material you give it (PDFs, slides, Word documents or a YouTube lecture), checks every answer against that material, drops questions that fail the check, and shows the sentence from your file that backs each answer, which makes the "check against the source" step quick. Our PDF to flashcards page shows how.
Frequently asked questions
Are AI flashcards as good as handwritten ones?
They can be, if they come from your own material, are accurate, and you test yourself on them repeatedly. You lose some of the benefit of writing them yourself, which you can win back by writing cards for the ones you miss.
What is the generation effect?
The finding that you remember information better when you produce it yourself than when you read it. A meta-analysis of 86 studies found a moderate average benefit.
Is it a waste of time to make flashcards?
No, but it becomes one if making them takes so long that you rarely review them. Time spent answering cards over several days is where most of the benefit comes from.
Should I handwrite or type my flashcards?
Either works. What matters more is that each card makes you recall something, and that you come back to the deck several times.
Sources
- Bertsch, S., Pesta, B. J., Wiscott, R., & McDaniel, M. A. (2007). The generation effect: A meta-analytic review. Memory & Cognition, 35(2), 201–210.
- Ebersbach, M., Feierabend, M., & Nazari, K. (2020). Comparing the effects of generating questions, testing, and restudying on students' long-term recall in university learning. Applied Cognitive Psychology, 34(3), 724–736.
- Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255.
- Kornell, N. (2009). Optimising learning using flashcards: Spacing is more effective than cramming. Applied Cognitive Psychology, 23(9), 1297–1317.
