I’ve only used an Echo and I’m definitely interested in hearing how the user experience of the Google Home is better. How is it more conversational?
From an AI practitioner: this is god-mode level work.
I've noticed no one my age clicks this. My son and his friends as well as a friends nieces and nephews who live a couple thousand miles away from us use Google in a completely different way. They don't go to google.com and type in a search query, they click the microphone and ask google their question directly. They are also more likely to use the assistant on their devices than any of the adults I know. I think this is one of those generational shifts as they don't need to be able to type in their queries in the best way to get the results they are looking for, they can just use natural language for the most part.
https://www.google.com/search?q=what+fraction+of+google+sear...
"What is one plus one?" "2"
"Plus one?" <Random movie reference>
"Add one to that?" "I don't understand"
"Add that to one?" "3"
Feels like an old text adventure.
It is almost always somewhere between faster and much faster to pull out my phone, switch app, type, than to waken up the device, ask it, wait for an excuse, rephrase my question and wait for it to read it.
That said it shows promise and if I still trusted Google I'd be somewhat enthusiastic :-| (Now I'm mostly scared.)
Since I do not trust them I have it in my office and turn the mic off when I'm not experimenting with it.
Next up: figure out if that mic is actually disabled in hardware if I "turn it off".
@Googlers: a good step forward wrt trust would be to allow the semi paranoid ones like me to turn on the mic on my phone on demand. I can live with the annoyance of tapping some button three times or whatever.
After China and a bunch of other stuff I do not trust you to value your relationship with your customers over your relationship with your bank.
Workflow one is leaving it on a desk and having it listen for the hotword all the time. No "screen must be unlocked" or "phone must be charging" nonsense; if I have to come to the desk and fiddle with the phone, I'll do what I need faster with my fingers. Workflow two is always listening when I'm on a Bluetooth headset. I can accept pressing a button on the headset to turn it on, but again, if I have to fiddle with my phone to get it to listen, it defeats the entire purpose of the tool.
(I'm also an old curmudgeon that is convinced speech recognition can be done entirely off-line, and there's no direct reason to send voice data to the mothership. But I guess it's fighting with windmills at this point.)