Will natural language be a predominate form of interface in the near future? Absolutely yes. But will it replace majority of interface? I am doubtful.
The reason for my doubt is that if you look around yourself; you will see a variety of amazing human computer and human machine interfaces, because no single interface has been universally suitable and more importantly the most or even reasonably efficient for all use cases. There is little reason to believe human language is an exception.
Switches, buttons, knobs, joysticks, motion sensors, and an array of other interfaces will persist until brain-computer interface is ubiquitous.
This is sort of a "what's a computer" question. I find that for a number of specialized digital interfaces, I don't want natural language. I don't want to stand there and talk to my microwave, I just want to punch in the number. I'd say about half the time I sit down at my TV I don't know what I want to watch, I want to scroll through some options. It feels like the more specialized the purpose, the less likely I am to want to talk to it.
> It feels like the more specialized the purpose, the less likely I am to want to talk to it.
This is the essence of it; and it follows not just for "shouldn't be a computer" appliances but also software. People in tech tend of to forget about the array of hyper-specialist software that exists in the wild, from little point of sales to industrial machines.
I have little interest in talking to a coffee machine or automated transport ticket machine. Flat White. Day ticket. That would do, thanks.
Uh, I might have a language barrier here. Doesn't predominate imply a majority?
But besides this I do see your point. Mechanical interaction with buttons, switches, dials and the like offer a great interface and they haven't been universally replaced by touch screens for a reason.
I was thinking more of the ubiquity of specialised phone apps. I checked my phone just now and I have apps for public transport (i.e. train tickets and routes), taxis/ubers, multiple apps for ordering food, fitness/workout tracking, todo apps, calendars, smart home control, weather, general search, news and lots more.
I'd say that most uses of most of those apps would be more convenient through a single chat like interface.
The reason for my doubt is that if you look around yourself; you will see a variety of amazing human computer and human machine interfaces, because no single interface has been universally suitable and more importantly the most or even reasonably efficient for all use cases. There is little reason to believe human language is an exception.
Switches, buttons, knobs, joysticks, motion sensors, and an array of other interfaces will persist until brain-computer interface is ubiquitous.