Imagine having AI translate whole paragraphs instead of just individual sentences. Then being forced to translate those paragraphs from English into target language. I know Luca Lampariello talks about this a lot. LingQs strength is being able to do something on large scales. And I think doing this with a percentage counter of how much matched the original text would be ideal. Because you would be forced to see what you're struggling to translte, think in target language, and if it focuses on a closeness accuracy, minor translation differences won't be negative. Getting 90% or above on your translation equal to the original, if we say you translate a 100 words would be very very rewarding.