This devices should by law always have a big LED on when recording, you would have the option to set it's brightness though.
Also why could it send data out without any confirmation from the user.
Also why could it send data out without any confirmation from the user.
With that said, my Sonos Alexa has a tiny LED but does respond with an audible tone when the keyword is heard.
That is... when it's not randomly laughing https://www.theverge.com/circuitbreaker/2018/3/7/17092334/am...
That would require, though, that it's not buffering the last N seconds of audio to reprocess once that processor wakes up. Do any/all of the modern smart-speaker devices do that? If so, then you'd have to take into account that when you see the light, you've potentially leaked any secrets you said in the last N seconds as well. Less like a reporter coming in and asking to speak to you; more like an eavesdropper coming in and telling you they heard what you were just saying through the door.
If, even with that expectation, enough people still use "voice assistant" technology for a judge to consider the tech to be "in general public use"[1], the bright line test defined by Kyllo v United States is triggered and the police no longer need a warrant to use the technology (in the abstract - they don['t need to use your hardware) to view "details of a private home that would previously have been unknowable without physical intrusion"[2].
Normalizing the expectation that previously private areas (like the inside of your home) might be recorded and sent to a remote 3rd party will eventually result in everyone losing some of their 4th amendment protection against search and seizure.
[1] http://caselaw.findlaw.com/us-supreme-court/533/27.html
[2] Ibid.
It's like the misguided Prop 65 signs that are in practically every commercial building in California so instead of warning people about dangerous levels of hazardous chemicals, they get a meaningless sign in every building for hazards that are no worse than if they were walking down the street or sitting at home.
As nerds, it makes sense to us to do local wake recognition and only upload a request phrase after that. But in reality, this not the first time[1] we've learned that continuous speech was getting uploaded.
Charitably, it could be a mistake, maybe the previous request never terminated and it just kept streaming. But again, not the first time.
More likely, my opinion, is there's a mountain of cash on the table in the form of private conversation, and AMZ would be foolish to walk away from it. After you talk about "hardwood floors" in your house, you could expect targeted ads, consumer profiling, variable pricing, and third party sales. Why woudln't they do this? Why else are they pushing home devices so hard? They're front and center on the front AMZ page every time you visit, aren't they? Oh and GOOG too, they just haven't been caught yet.
1. https://www.wired.com/2017/02/murder-case-tests-alexas-devot...
And the LED should be forbidden by law from being software-controlled.
They should also have a physical inline toggle switch to disconnect the microphone(s) entirely, whose state is easy to visually confirm.
We're also discussing theoretical hardware changes, there's no reason it can't have 2 sets of microphones where one is hardwired only to the wake-word processor which has no direct connection to the main processor except some one-way signalling, and the LED is hardwired to the second set.
The idea that there would somehow be a dedicated wake word microphone is a little ridiculous. Firstly, no one would trust this supposed 1-way connection. Second, it would require a dedicated processor to make the wake word even work, driving up costs. Third, the echo uses an array of microphones so your wake word would either be unreliable or drive costs up further as you duplicate the entire array. Hardly a net win.
The reality is that if you don’t trust amazon to do the right thing you shouldn’t install their listening device in your home. (Likewise for your phone.)
WiFi is always on. WiFi doesn’t disconnect just because you aren’t actively moving data. You’d be pretty annoyed if every time you wanted to send or receive data your device had to reconnect.
If the microphone uses an amplifier, you could wire an LED to light when power is supplied to the amp. The indicator light is physically part of the circuit, so its operation cannot be modified by software. There are probably other, better, ways to do it, I'm not an electronics guy.
I don't think that would be a bad thing; especially if there's a switch to disable the mic. When you turn it off, you'll get reliable feedback to know it's actually off.
I think it's important that these kinds of devices have simple feedback and control mechanisms that can be independently verified and reasoned about. Software is too opaque and too untrustworthy.
I see that, I think it would be best as a separate light.
I think product UX has drifted too far towards blank monoliths; I know I wouldn't mind a few more blinkenlights :)
Doing it via microphone/amp wouldn't do it, since you'd still want to use it without LED on (and not software controlled).
It has that. They call it the power cable.
If you disconnect the microphone, an echo is pretty useless.
We don't live in a world with a healthy enough market or enough competition for that to work.
Talking loudly in another room. Alexa: "Do you want to send this message to Jim?" "Alexa, you suck"
A confirmation after every request makes the device a lot less useful, I suspect most people do not want the device to confirm commands before it does them.