I thought about input interface that could replace keyboard for VR and mobile devices. Brain implant would be perfect, but that's not feasible with today's technology. Voice is good contender, except for privacy issues, and it's not really usable in work environments. Keyboards are our best solution so far, but they don't evolve and to me they seem like a dead end (plus, they are not portable and require both hands to use effectively).
I was thinking about touch sensitive surface that recognizes drawn glyphs. The idea is that anyone who is literate can start using it without any training, just draw letters instead of typing them. With machine learning and some clever visual feedback, both user and machine could adapt to each other to increase the input speed (by simplifying glyphs and by defining 'snippets'). Interface could be expanded to both hands to double the speed, but it would be completely functional with just one hand. It could even be used blindly. For mobile devices, such sensitive surface could be placed on backside - virtual keyboards that take half of screen are just horrible.
