The interesting thing here is that nobody writes chinese one character at a time with pinyin; you almost always type out an entire phrase in pinyin, and usually there's only one meaningfully correct combination of those sounds in terms of characters and meaning (that's how the listener can tell what you're saying, after all, when listening) which will be the first choice in your input software (input software traditionally gives you a first choice with what it thinks the entire phrase is, and gets shorter with the 2nd choice onwards to partially match a phrase.)

The problem is not that people can't recognise the words; it's that we can't write them if given pen a paper. If the phone gets it wrong you just choose the nth choice instead of the first.