Yes. The computer should never do something that I haven't explicitly told it to do. You can design the same exact feature but I need to trigger it, the machine should not be making decisions for me.
Because the OS can guess correctly more often than a typical user can type correctly. You can always proof read its work at the end.
Also, in addition to auto correcting words, your phone also auto corrects individual key presses based on the word you are typing. The same physical impact on your screen will register a different letter based on the letters that come before it.
So part of your “typing on glass accuracy” is dependent on a designer designing that feature that “does work ahead of time”.
> Because the OS can guess correctly more often than a typical user can type correctly. You can always proof read its work at the end.
No. Certainly not on a desktop, where I type much faster than any suggestion can appear (esp. infuriating on IDEs); usually not on a phone either, where "corrections", while sometimes faster than me, are always wrong.
I use SwiftKey (without cloud predictions) and if you press spacebar that's the equivalent of "explicitly telling it". The thing is that usually, the auto correct is correct so when you see it is not you click on the word you (according to the computer) "made up", et voila.
