I find speaking to computers (say Alexa, Siri, Cortana, …) to be tiring compared to speaking with people and slower than typing.
Typing in VR with visual tracking or controllers is practical if you need to log into something or bash out a Tweet, but you want to get a real keyboard aligned with your space if you are going to type much.
The myoelectric device that was used with Meta’s Orion demo could lead to something better, you can get signals from a EEG or my electric array and train it and train the human to communicate in symbols across it. With a direct brain interface, do any better.
Some of my distinct style as an HN commenter comes from my perpetual fight with my iPad. I am sitting in the couch holding it in landscape mode with two hands and typing with just two thumbs. I can do it crazy fast. The text input system has a predictive model which catches and fixes many mistakes that I make (don’t want to turn it off) but also injects errors of its own. (It just did it.). I don’t catch all of these so you find grammatically probable but semantically wrong errors all over what I write.