I turned the autocorrect off on mine because it kept making incorrect corrections. I suspect a bunch of ESL DEI hires are currently employed maintaining those spellcheck features.
It's not just the phones, something happened to spell-check in browsers too.
I've seen Chrome-based browsers not able to correct one character wrong in a twelve character long word, not even needing to add or remove a character just fix one vowel. A 1980s spell checker using 20k of memory was better than what we have today.
I haven't looked into why but I'd guess when they 'upgraded' it for unicode and non-English languages the programmers assigned to it were idiots. Eventually they'll dumb it down even more and use a hundred billion parameter LLM to do it as well as Levenshtein and a Trie.
I turned the autocorrect off on mine because it kept making incorrect corrections. I suspect a bunch of ESL DEI hires are currently employed maintaining those spellcheck features.
It's not just the phones, something happened to spell-check in browsers too.
I've seen Chrome-based browsers not able to correct one character wrong in a twelve character long word, not even needing to add or remove a character just fix one vowel. A 1980s spell checker using 20k of memory was better than what we have today.
I haven't looked into why but I'd guess when they 'upgraded' it for unicode and non-English languages the programmers assigned to it were idiots. Eventually they'll dumb it down even more and use a hundred billion parameter LLM to do it as well as Levenshtein and a Trie.
edit: Yep. Chrome and Firefox now use Hunspell and "Hunspell [is] designed for languages with rich morphology and complex word compounding ... Unicode UTF-8-encoded". Non-CS degree lead programmer on it brags about non-standard hyphenation support /facepalm