Re: I could train 1st line to be fluent in 'user'
I have something like this. I have to carefully re-read pretty much everything I write as the first draft is likely to contain extra, missing or transposed words. And I see it quite a lot in other (technical) colleagues replies to questions, where they are missing a crucial "not" that would normally alter the sense of the answer to the opposite of what was intended. But from the context of the question, and the phrasing of the answer, you can see the missing "not". I don't know if any AI systems are being trained to do the same.