Rather than the cow pat that Apple AI is, I'd love some sort of ML model that could fix parts of dictated text that sound similar to what was intended but actually make no sense. A recent example would be dictation coming up with "on styled" (which was grammatically incorrect in the specific context) rather than "unstyled" which could've been inferred from the meaning of the overall sentence.
=> More informations about this toot | View the thread | More toots from jscholes@dragonscave.space
text/geminiThis content has been proxied by September (3851b).