Siri, Alexa, Google Home/Now, Cortana… One thing none of those links gave you was the ability to use the assistant from your web-browser. You have to install the right app for a specific ecosystem on specific hardware :(
A family household is likely to be a complicated device-ecosystem, a typical house is going to have TVs, computers or laptops, tablets or pads, phones, music and/or game systems.
I want you to try role-playing this scenario, ideally sitting down.
You have a household assistant, you can ask it who is at the door, get it to place phone/skype calls, change the tv channels, play music… You just have to say “Hey house” and the command you want.
Ok. It’s cold. Ask the house to “turn up the heat”.
“Hey house: turn up the heat”.
Great. The room starts to warm up.
Now, you can go back to reading on your tablet/laptop/phone this great news that your bank noticed it hasn’t been paying you interest and you’ll see a $1,000 deposit from them in your account tomorrow. There has to be a catch, or some small print, right? But it’s a bit dark and you can’t read the screen. There are no brightness controls on this thing, you have to ask the house to raise the brightness on the display.
Go ahead. Ask.
If you’ve used a few of these things, you probably said “hey house please turn up the brightness on my tablet display” :) Gah.
Non-initiates probably said something more like “hey house turn up the brightness”. For this to work, the house has to know which display you mean’t, and the chances are that you looked up/away from the device when you spoke. So even if the house was paying attention, it couldn’t work that out.
I think the solution is to have one system with many identities, instead of this Siri/Alexa/Cortana approach hardware vendors are taking.
The way I see it, your devices should work as a team. Sure, you might have an overall system that has a name, so you can say “House, turn up the lights” or “House, turn the brightness down on the tv”, but if we’re going to include these devices in our daily verbal communications, they’re going to need naming. It’s no sillier than giving a domesticated canine or gold fish a name. It’s how humans work. Things that get spoken to need names.
Those names could just be “phone”, “kindle”, or “shitty laptop”, I’m not saying you have to make long-term commitment life names like “billy joe ray” for your blue ray player (well, actually, that’s not that terrible, I just wouldn’t want to have to keep saying that).
The goal here should be that I can speak to/about any of the major devices in my house through a singular system.
I should be able to speak to/of the TV, the xbox, the PC in my office, the kindle, any of my devices, through a singular assistant, and customize the experience or specialties of any of them, perhaps through a common interface.
Individual devices might have a pretty limited vocabulary, and that’s OK when they’re all working through a common, shared, interface.
But when the device itself is the driver for such an assistant, it can feel cold and impersonal in a way that doesn’t foster verbal dialog.
Cortana, for example, feels like the most basic and crude of any of the assistants I’ve used. It has no charm or warmth to it, and outside of tablet mode – e.g. on a surface or a PC – it’s abominably intrusive.
The problem is that Cortana is behaving like both an assistant and a discrete UI. When I’m using the device, I don’t want that, and that’s a problem with the assistant being a part of the device like this.
What I’m envisioning is a sort of master assistant that knows how to forward requests. So when I say “Hey, PC, play me some clapton”, my primary assistant actually handles the request and turns it into some sort of inter-ai message that gets the PC to do what I want.
This would help to mask the limited capabilities of individual devices, while still allowing you to develop a sense of how to communicate with a given device.
But right now, each individual assistant is an island, and that’s really going to become a huge pain going forward.
It’s a great time for someone to step in with a meta-assisstant.