A related concept to "autocorrect" (fix it after typing it) is "predictive typing" (expand invisible hit zone around statistically likely onscreen key before the user presses it).
This was the clever technique that Apple programmer Ken Kocienda[1] figured out for the 2007 iPhone to make its tiny QWERTY screen keyboard usable.
No. Intellisense tells you a list of options based on type inspection and then optionally autocompletes from the set of options. It just says that if you type A and then a dot, according to the grammar of the language, you then must be about to type a member of A. But it's not predictive, because you literally just told the system that you're about to type a member of A by typing A and then a dot.
> Don't you also get predictive variable and function names? Types? Control structures? enumerations?
Nothing* in intellisense is predictive in any meaningful sense. It just displays the entire set of options constrained by one or more of [the grammar, structured object inspection, and words seen in the document].
* There is one thing, actually. Intellisense can optionally boost the priority of a suggestion based on proximity to other instances. I would classify that as prediction.
I'm not surprised that autocorrect came out of Microsoft. The best autocorrect I have ever used was on the keyboard of the Microsoft Zune HD. If you were even hitting the correct half of the keyboard, it would be able to solve the puzzle and correct your typing. It is the only onscreen keyboard that even comes close to using a physical keyboard on a phone, and I'm routinely disappointed by Apple and Google's comparatively pathetic offerings.
Windows Phone 8 (or even 10) with wordflow was the best prediction/autocorrect experience I've used. I sometimes want to fling my iPhone at a wall in frustration.
My only experience with android is a low-powered phone and Swype on that was quite bad, but maybe it's better on high-end Androids ? Or is there a better keyboard over stock and swype ?
For those interested in AI: the root of autocorrect appears to lie with Warren Titelman (also originator of things like Undo) who was a PhD student of Marvin Minsky at the MIT AI lab. Warren invented DWIM -- Do What I Mean, like an autocorrect for the shell, when he worked on Interlisp at PARC. Simonyi, author of the first WYSIWYG editor, Bravo, left PARC to go to Microsoft and re-implement Bravo as Word, incorporating a lot of other PARC ideas that were in his head.
DWIM was a bit weird and as I was used to typing what I wanted I didn't really like it (though strangely, macro expansion was tied into DWIM so you couldn't disable it completely). Though it was a lisp thing (and Interlisp-D wasn't even written using a text editor) the best way of thinking about it is if you typed 'mr -R /' into Bash and bash suggested both that you meant "rm -r /" and that doing so on the root would be a bad idea.
> Without it, we probably couldn't even have phones that look anything like the ingots we tickle—the whole notion of touchscreen typing, where our podgy physical fingers are expected to land with precision on tiny virtual keys, is viable only when we have some serious software to tidy up after us.
This is a nice dramatic sounding intro, but it is really exaggerated. Along the lines of "Nobody can survive without social media in the modern world". Of course you can live without it. Stop trying to inflate a minor inconvenience that you can get used to to "impossible".
I never had autocorrect and I recently found the option to switch of the "suggestions" in the top bar of my phone keyboard. It is heaven. Sure, I have to spend some time fixing my typos, but it's way faster than to fix a completely wrong suggested word that I accidentally touched.
This was the clever technique that Apple programmer Ken Kocienda[1] figured out for the 2007 iPhone to make its tiny QWERTY screen keyboard usable.
Deep link of Scott Forstall explaining it: https://youtu.be/xxBc1c3uAJw?t=4m55s
[1] https://www.wired.com/story/opinion-i-invented-autocorrect/