Comment by kqr
1 day ago
> There are almost no standalone tools dedicated to creating flashcards easily from existing programs
I think this is a common misunderstanding. Half the benefit of SRS comes from working out what the flashcards are. You have to circle around a concept, look for similarities, differences, examples, generalisations, properties, etc.
Is it hard work? Yes. Does it help understanding? Massively.
This is also a very difficult skill which, I believe, is why many people fail to appreciate SRS. They try, write bad flashcards, don't see results, and give up.
EDIT: This also leads to another common misunderstanding, that SRS is only good for memorising facts. With proper elaboration (thanks child comment), it can be used to build understanding of complex subjects too.
> Half the benefit of SRS comes from working out what the flashcards are. You have to circle around a concept, look for similarities, differences, examples, generalisations, properties, etc.
This is at least half true, but among SRS folks it's overstated.
The same argument stands for any creative endeavor.
Debussy had a greater experience than I did with Claire de Lune, because he composed it. In turn, my experience of it is probably greater than the next persons' because I learned it. Bottom of the heap is the lowly listener - but I think everyone would agree that listening is a wonderful way to appreciate music and shouldn't be diminished.
Same for literature, visual arts, etc etc.
The still greater reason that create-your-own has historically dominated for SRS decks is the specific curation aspect. Anything curated by someone else suffers from mismatch with your own intentions and prior knowledge profile.
(Plug time. Like most people, I am building my own grandiose SRS app. Some writing on it is here: http://patched.network )
This looks amazing. I love your description
> …the FOSS lovechild of anki, duolingo, wikipedia, and MathAcadamy, with a more generalized surface area…
I haven’t read much in your blog and the answers are probably there, but what’s the current state? It looks like you’ve done a lot of work on it. Is it in a usable state at this point?
Wow, thanks!
The app's been 'in production' for myself and my daughters for years, but no public access or polished working demos yet. I hope to have exemplar static-site builds using very soon.
I'm also working on these guys for my in-development early literacy course: https://patched-network.github.io/letterpeople/
Creating your own flashcards and choosing what to curate is a helpful part of the process. This doesn't mean that being able to snip something nicely into a flashcard is not beneficial.
You might be right. I remember when I used SRS seriously for language learning, I got far more out of the cards I created painstakingly than the pre-built ones that I pulled off the shelf. That is - I still remember most if not all of the custom cards verbatim.
> They try, write bad flashcards, don't see results, and give up.
But if that's the case, then wouldn't a program that takes a long text (like a book) and creates GOOD flashcards, be way superior over someone making their own bad flashcards?
The quality of a card isn't absolute, it's personalized to the learner. There are ways to make an almost universally bad card (e.g. putting way too much information or a misleading image), but a really "good" card comes from the learner engaging with the material first before making the card.
The program doesn't know what's in the learner's head and what associations the learner will make. That's the benefit of making your own card.
I can see that helping a beginner to get a lay of the land, but I see that as no more than training wheels.
I created bad flashcards for a long time. In the end, I just figured, constantly editing cards when I noticed issues was the best way to learn how to use flashcards properly.
I have just tried this afternoon to create with Gemini 2.5 Pro Anki cards to study for my exams. I've been doing it raw: I just paste the whole material (like 100k worth of tokens) into aistudio and generate the flashcards in txt format.
For now it's going great, and I think that the cards are good enough as they are (there might be a couple here and there that I would remove or edit, but they don't get in the way of learning).
Is there any prompt tips you have learned on that path, that you could share? or the obvious request has been useful already without too much fiddling?
I was thinking to use some study material to test drive Google NotebookLM for this use case.
2 replies →
Or a person could have the program either critique their flashcards as they write them, or suggest new sorts of flashcards to create without doing the work for them by automatically generating them.
I do this with my kids' "exam review" materials. I'll take a photo of the pages marked "study this for test" and send to Gemini and say "make a test out of this" and it'll create an exam with questions, answers, multiple choice, true/false, fill in the blank, etc.
It's really impressive. So I just tried it to do what you were asking:
> Give me the first three paragraphs of Genji Monogatari in Japanese (I've only done this for an initial seed text; you'd probably instead use your own source text)
Here are the first three paragraphs of Genji Monogatari (源氏物語) in Japanese:
いづれの御時にか、[snip]
> Create spaced learning repetition of those three paragraphs using cloze detection for any word that is not in the top 5000 most common Japanese words
Okay, I can help you create a spaced repetition exercise for those three paragraphs from Genji Monogatari. This will use cloze deletion (fill-in-the-blanks) for words that are generally less common in modern Japanese or are characteristic of classical Japanese. Since I don't have a real-time tool to check against a specific "top 5000 words" list, I'll select words based on common knowledge of Japanese vocabulary and classical forms that would typically be targeted for study.
Please note that the third paragraph you provided contained "тревоしき", which appears to be a transcription error. [This is true. Gemini's first three paras of this novel did contain this transcription error.] I will substitute it with "あさましき" (asamashiki), meaning "shocking" or "deplorable," which fits the context of referencing Yang Guifei and potential societal disruption. If you believe another word was intended, let me know!
What I received was a canvas I can export to Docs, and it begins with:
Genji Monogatari - Spaced Repetition Cloze Exercise This exercise is designed to help you learn and recall less common vocabulary and classical expressions from the beginning of Genji Monogatari.
Paragraph 1 - Cloze Test いづれの ______ にか、______ 、______ あまた ______ たまひけるなかに、いと ______ き際にはあらぬが、すぐれて ______ たまふあり ______ 。はじめより我はと ______ たまへる ______ 、______ ものに ______ そねみたまふ。同じほど、それより ______ の更衣たちは、ましてやすからず。
Paragraph 1 - Answer Key いづれの 御時（おんとき）にか、女御（にょうご）、更衣（こうい）あまた さぶらひたまひけるなかに、いと やむごとなき（やんごとなき）際にはあらぬが、すぐれて 時めき（ときめき）たまふありけり。はじめより我はと 思ひあがり（おもいあがり）たまへる 御方々（おんかたがた）、めざましきものに おとしめ そねみたまふ。同じほど、それより 下﨟（げろう）の更衣たちは、ましてやすからず。
Paragraph 1 - Vocabulary 御時 (おんとき): reign, era (honorific) 女御 (にょうご): Imperial consort of high rank 更衣 (こうい): Imperial consort of lower rank than Nyōgo あまた (amata): many, much (classical) さぶらひ (saburahi): (archaic form of 侍う - saburau) to serve, to be in attendance (humble) やむごとなき (yamugotonaki/yangotonaki): exalted, of very high rank, distinguished (classical)
[it continues the same way with para 2, 3]
How many errors were there in the final result?
>I think this is a common misunderstanding. Half the benefit of SRS comes from working out what the flashcards are. You have to circle around a concept, look for similarities, differences, examples, generalisations, properties, etc.
Yes. I really get the urge to automate stuff, but imo half of the benefit is from doing the leg work.
The technical term is "elaboration" and it is key to long term retention.
Fair point, please see my comment in response to allenu, in this same thread, for a detailed view.
I did not mean to glaze over this aspect, I am aware this is very important. I can not and do not use AI for automating flashcards for this reason alone. But, I think my point still stands, getting information from one disparate app into another adds a lot of friction (saving it somewhere, copy and pasting and yada yada) to something that (I believe) should be as easy as possible. We have 1 click instant payment, but I can't have 1 click get this into some sort of inbox in Anki ready to flashcard-ify?
To be honest, I have a whole frustration with apps and windows that I'm still trying to word, but fundamentally all we are doing when we are "computing" is moving information around. I wish that information was a first class citizen at the OS level that could be leveraged by any app immediately. Utopian view but this stuff ain't gonna think about itself.
That's usually done by writing and elaborating notes. I think it's not clear to most people how flash cards improve this. The original idea of flash cards is rote memorization of facts that need fast (automatic) retrieval.
Agree - making cards yourself is like programming yourself for how you want to understand and remember an idea.
I grabbed a set of flashcard for Spanish verb conjugations. The guy had put in a lot of work to get them in the most frequent order, over 3000 lines of python. It been a massive help to me and I am not sure that writing the code to generate the list would really have helped my Spanish too much.
https://www.asiteaboutnothing.net/w_ultimate_spanish_conjuga...