"A cat is bigger than a mouse."
Translation:Eine Katze ist größer als eine Maus.
This would have been shown by Duo as the 'correct' answer if someone wrote 'ein Katze'. Duo sees the masculine article doesn't match the feminine noun, and so suggests an acceptable masculine noun. It would probably make more sense to correct the article, but this is what we've got.
Indeed. A never-ending source of confusion to learners.
I'm not sure whether it's possible to get this right without making the correction algorithm language-aware so that it knows what is more likely to be wrong (e.g. article gender or adjective case ending).
That doesn't seem like an unreasonable goal for a language-learning platform, to be honest.
Of course. It's just more work (= more programmer time and money) than the simple version, and doesn't happen by itself.
You'd also have to implement it (or at least parts of it) separately for each language rather than having "one size fits all".