Voice-responsive AI assistants are all over the place now, but do any of them make allowances for the (perhaps unlikely) situation where you need to frequently say something that sounds like their name? Do any of them allow you to rename them?
They don’t allow you to rename them since the engineers pick things that are supposed to be easy for the software to recognize and spent a lot of time improving the recognition around that phrase. If your pet had that name, you would most likely constantly engage the voice recognition mode.
There was an amusing case of this back when Microsoft launched the Xbox One, which you could power on with a voice command. They showed an actor turning on his Xbox console with the voice recognition and consoles at home were picking up the same command. At least, that’s the story.
You can always rename your dog.
My next door neighbor’s dog is named Bo.
Have to stop and think before I try and redirect him.
To continue the hijack (sorry), one of MY dogs is named Basso after Ivan Basso.
You can guess what that sounds like when I yell his name to call him.
If you think your dog has fleas, you have to do a search on Google.
I’ve heard of one family (friends of my sister’s) who already had a daughter named Alexa when they got the device. You try telling your daughter that you’re changing her name for the benefit of some machine.
These things have to be able to recognize enough different words as it is, that if they needed that level of individual fine-tuning, they’d be useless.
Google voice command only works on my phone if I say the specific phrase “OK, Google.” If my dog were named Google and I just said that name without the “OK,” my phone wouldn’t respond. So I could call “Come here, Google,” or say, “Sit, Google,” or say, “Speak, Google,” and only the dog would respond.
If your puppy is Greek, and you are trying to teach it to not do things, and you say “όχι, Google” and popping its nose with a rolled up newspaper, it will probably summon a voice search engine.
Amazon has addressed this.
You can rename Alexa to Amazon if you’d prefer, but that’s it for now.
I wonder how other languages deal with it? “Siri” is probably generic enough to work across different languages, but “Alexa” isn’t exactly easy to say and Google has distinctively different pronunciations in other parts of the world.
…or Syndrome.
One of my daughter’s children is named Alexis. When they bought the Amazon Echo (which by default answers to the name “Alexa”), they changed its response phrase to “Echo”.
(I believe they were limited to one of three choices: “Alexa”, “Amazon” or “Echo”.)
A local radio DJ had a ‘dog’ named “Stain”. On air, he would call the dog to approach him; I’ll let you figure out what word he used to call that dog.
An Amazon Echo commercial came on once while I was watching tv.
Girl in the commercial said “Alexa” about ten times commanding it to do all sorts of things.
As predicted, MY Echo went batshit crazy trying to keep up with all the command prompts.
ROFL!
“Come, Stain!”
As I understand it, phones that are always listening have some dedicated low-power hardware. It’s just good enough to recognize “Alexa, …” or “OK Google, …” and is optimized for the task. Subsequent general-purpose voice recognition can use the main CPU on the phone, and send parts of the recording to Apple/Google/Amazon’s servers for further processing if necessary. If the phone used the general-purpose voice recognition all the time to listen for the triggering word, your phone would quickly drain its battery.
My WAG is that the always-on listening hardware isn’t completely hardcoded for a specific trigger word, but rather is optimized for easily recognized phonemes. If that’s the case, the user could specify other trigger words, but I imagine it wouldn’t be easy to come up with a straightforward description of what makes a good trigger word. That’s a recipe for user frustration: “Stupid piece of shit doesn’t recognize ‘Quetzalcoatl’!”
There’s definitely some intelligence built in to it, because you can train it to recognize your voiceprint and use that to unlock your phone. Someone else saying OK Google while it’s locked won’t unlock your phone, even though they can use it just fine after it’s unlocked.
However, the search set is much, much smaller. The incoming recording just has to match (to some %) either the words “OK Google” or a trained voiceprint. It doesn’t have to try to guess, based on grammar and context, what you’re trying to say. And it doesn’t have to send that over the Internet to server farms the way the normal voice recognition works.
I would think that there would be a low level listening, but then that data would be sent on to a higher level function to verify. The low level listen has to be there to avoid killing the battery, but it’s just a first approximation.
My Xbox One occasionally misinterprets things it hears as commands. It can be annoying. That’s just a downside of the technology.
I’m glad my toddler isn’t named Xbox or I’d have to unplug my Kinect permanently. 
Yeah, I guess since humans do that too (“what? Oh, sorry, I thought you called my name…”), it’s going to be a natural side effect of having properly implemented voice recognition.
There was a thing a while back where some players in multiplayer XBox games were changing their public names to things like “XBox Sign Out” - when another player encountered them and said (out loud) “Dude, why are you called XBox Sign Out?”, they would be thrown out of the game as their system shut down.