I think there's a character (or google thinks there's a character) after "paragraph" that it can't parse so the code sees it as undefined, and is then being converted into text for spell check literally as "undefined" so it's checking "paragraphundefined." probably because of AI shenanigans
It’s *probably* not AI except in the sense that all automated systems are AI, but yup this is the answer. It also looks like it’s converting whatever null character to text before spell checking so it’s checking “paragraphundefined” which is where it’s getting the split or hyphen it from.
There’s not really an agreed upon definition of what counts as “AI” so I’ve seen it commonly used to mean anything from “any computerized decision making system” to “LLMs and image generation”
While some grammar check systems are more sophisticated than classic spell check and I’m sure there are ones incorporating LLMs now there’s not any reason to think this is one of those. This is a very classic type of mistake a traditional spell check system would make
I imagine we essentially taught our autocorrect systems then much like we teach our AI/LLMs today but more deliberately given the time and specific task at hand. This makes me think that it’s absolutely possible it’s gotten worse if we never stopped the learning process.
Particularly given the ways that written communication has changed since the introduction of autocorrect. Or maybe auto correct is just something that will go through cycles of being good and bad as it constantly attempts to keep up with our ever evolving languages.
Comments
(these are all real Dutch words that Google Docs has marked wrong at least one time, and yes that includes the country of Peru)
(and I see also "alinea", which is the Dutch word for paragraph)
also i swear spellcheck/grammar check programs are usually some level of ML (so more than just routine automation)
this.