Our beloved SaaSodile is under attack! While I’m sceptical of reports of an imminent SaaSpocalypse, I do see a bleaker future for naive crustaceans with big dreams.
WaniKani charged $9/month to teach you that 木 means tree, which Claude will now do for free while also explaining the entire history of Japanese forestry.
Cheeky. Claude definitely ain’t free and also highlights how much crap LLMs can generate. But it’s also a pretty powerful tool which gives you anything from a chat-based personal tutor to… making your own SRS learning app
What do you think? Will WK (et al) be consumed by sycophantic know-it-all chatbots or will the humble cost of server + database outlast the MASSIVE cost of token inference in this learning war of attrition?
AI could be exceptional at adding explanation and I still wouldn’t use it for something like this. They’re still insanely prone to just making up bullshit and spouting it confidently, and when I’m learning something like this, I don’t have the knowledge to know when it’s wrong, so I’m not going to trust that they aren’t just teaching the wrong things.
At most I would use AI to help craft a mnemonic that I can’t think of, or create an image for a mnemonic.
I’ve heard that you should use AI for supplementing your learning, not outright replacing it and I agree. If you use it for making a mnemonic or for practice quizzes, that’s completely fine imo. You shouldn’t learn new material because of its tendency to just hallucinate whatever
It doesn’t seem like AI is remotely relevant to the health of WaniKani when Anki exists.
If it’s worth it to a viable amount of people to offload the service of curation, direction, presentation, and management enough to not instead source and install an approximate alternative deck for a free, long-lived and reliable customizable tool, why on earth would it not be worth it vs. jamming words into a prompt until the same hopefully pops out from scratch?
The appeal has never been about “it’s impossible to do this another way,” rather it’s about being one specific appealing and trustworthy way with support and community built around it, which is nice to rely on when venturing into learning something you’re a complete beginner in.
AI might be able to generate the whole app similar to Anki, content aside. Well, even for content, it’s possible to avoid AI-generated content (use Jisho data sources, only use AI for coding algo) but maybe not Context Sentences and Pattern of Use…
It’s not like using AI to write grammar explanation. WK doesn’t teach grammar. I’d even argue that AI-gen sentences are doubtful.
That would be similar to Anki, but with more streamlined UI. AI coding, maybe plus AI mnemonics, that is. Honestly, I like WK algo more than Anki, but WK has lack of real reliable customization (unless using third-party mobile apps).
After all, it’s similar to what WK and Anki can do and cannot do. Even WK doesn’t teach the whole language, and might not work well for everyone.
Right now Gen AI’s biggest output seems to be articles talking about how great Gen AI is and how it will replace everything. I see much less actual real world adoption from people without financial motives
