There has always been the myth/rumor/belief that our devices are always listening. I wonder if this is an overton window moment, get people used to a device actually listening and slowly move to all devices listening.
I really wish these conversations would shift to the scarier truth: companies collect so much metadata on everyone that they can make pretty good guesses without processing what you're saying.
If you think "oh hey I was just talking about socks, how did they know to show me an ad for socks" it's likely because they accurately predicted based on your behaviors that you are highly likely to be in the market for socks right now.
None of the companies would have an incentive to report it.
Or you’re talking about socks because of the previous ad for socks you saw. I suspect ads drive our collective behaviour more than we realize.
I'm not kidding about how insidious this is... there's so much metadata to exploit that these companies know who you're with, when you're with them, how often you're there, etc they don't even need to explicitly ask for location because they can pick it up based on network info. In some cases they probably could tell you more about a person than that person could tell you themselves.
Whatever side you come down on, privacy doesn't stand a chance against a small amount of convenience.
‘Listening’ sort of implies intelligence is paying attention to what is said. It’s more ‘hearing’ than ‘listening’
There are documents on how these activation words work
- The microphone always vibrates from the sound waves
- The microphone is powered in a way that the sound waves change electrical readings in some way
- The electrical readings are sent to another component reading them
- A component receiving the data does some kind of unbuffered processing related to triggered actions (e.g. a clapper or a activation wave)
- Some component temporarily uses a buffer of the data but not for permanent storage (e.g. live, unstored transcription for the deaf or a 'nevermind' after a triggered activation)
- Some component stores or sends data generated by the sound, but not necessarily the original audio or even any attributions of who (e.g. voice trigger web search sends the search query as text)
- Some component generates a stored copy of the transcription with attribution of who
- Some component stores the actual audio in a way that can be later replayed
I'd say this "sounds" like a mess to deal with, but then I'd be worried about falling into a category ;).
What the smart speakers and devices do with the keyword is closer to the clapper than an actual transcription.
The tech is already included in the iPhone, and storage requirements are in the hundreds of Megabytes, not more. The device is a black box on purpose, and my informed opinion is that they have been logging the ambient text-to-speech since a long time.