What I mean is that if I entered a sentence into ChatGPT/Gemini/Grok and tell it to fix the flagged word, it will be able to get it right almost all of the time (assuming it's not a weird proper noun or inside joke slang).
What I mean is that if I entered a sentence into ChatGPT/Gemini/Grok and tell it to fix the flagged word, it will be able to get it right almost all of the time (assuming it's not a weird proper noun or inside joke slang).
Often, but not always. For my thesis, I ended up with a section related to porn. ChatGPT simply refused to spell-check that section. Also, recently I wrote a comment on HN about different subsets of English being easier to learn for native Spanish or German speakers, and the Samsung AI spellchecker refused to review it because it was considered "inappropriate content."