Next token predictors have largely solved this for day-to-day use.
Swipe + a more modern prediction model would mean we'd be able to type one-handed without looking at the screen at all, and have perfect accuracy.
We have the tech today. For whatever reason, nobody is using it. Why?
PaulHoule•2h ago
mvkel•1h ago
^
This is what appeared when I said "it's even worse"