(no title)
barnabyjones | 1 year ago
Personally I find Duolingo great for Japanese, I have the opposite problem: I run out of hearts if I try to do review and the app rarely gives me much old content, so I don't practice it. If this could come up with Duolingo-esque lessons for different levels I might use it to supplement. But letting users choose words/a topic is not necessarily important, most people don't know what they should study so it might be better to have "recommended" premade decks for different levels.
williamsss|1 year ago
>The AI coming up with words for you doesn't add a ton of value, most languages have standardized tests with good word lists that learners should already be using. Adding the issues others mentioned, I think this is feasible but needs agents taking additional steps to mimic what a real tutor would do and formulate a useful short lesson.
I plan to refine the content certainly. If you have examples I could use for a baseline to achieve what you want that would help me back test against I'll try to add that into the next release batch.
> If this could come up with Duolingo-esque lessons for different levels I might use it to supplement. I see the app to be used in tandem with a structured lesson rather than structured lessons in the app. Similar to an Anki study deck.
I had looked into lessons for the app on my 3rd iteration of this and went through a massive refactor for it all to be lesson based. In the end, it felt like AI slop so I kept it simple with vocab to start. In any case, I will refine the marketing copy to reflect that.