The most disruptive errors are those that could seemingly be cleaned up with some simple rules (e.g. if the word is in the dictionary, then don’t change it). This doesn’t seem like a “state of ML” issue—the Gmail text completion is annoyingly good, and you-know-who can compose soliloquies and smut with uncanny skill.

And no, I don’t often discuss The Rock, but if I did (“Do you smell what the r...”) I'd expect it to finish the alley-oop. I do not want to go Rock climbing.

iMessage is software that is used <Carl Sagan voice> billions and billions of times a day, and half the time it feels like verbal jiu-jitsu (“jiu-hursuit” according to autocorrect just now) with Dory the fish.

It isn't just you that affects the learning.

> if the word is in the dictionary, then don’t change it

Actually the thing that bugs me the most is that when there's a word that is in the dictionary that's long and complex and it doesn't suggest it, even when I have the right word. It just isn't underlined red. Or a one off error with these words and it has no idea what I'm trying to spell.

They have a long way to go. These systems aren't just ML btw.

Predicting subsequent words is hard for humans, but correcting typos is easy. I've turned off autocorrect completely because my typos are always immediately decodable to their human audience, moreso than Apple's seemingly random modifications.

Could autocorrect not wait until it sees more of the sentence?

Android autocorrect does this and it's maddening, because it's very bad at it, and now you don't notice the word it changed because you're looking at the subsequent one.

Having to monitor several words back to fix errors that the system adds is exhausting. I wish I could turn it all off, or at least revert to the system from the early days of iOS.