The new utility to get sentences and their improved version is available now.
Are there more or less real plans to use this input for training an AI to be part of LT?
I have marked a lot of Dutch sentences as containing an error. Lots of those are spelling errors, secondly it is compounding errors, and less so grammar and cases of word confusion (like hè, he and hé, all with a different meaning, but used in each others place).
I could start an effort, challenge, to have people correct these errors. But there is only use in it if plans for training an AI are real.