Ask HN: Should I give in to voice assistants? I've been resisting privacy-invading tech as much as I can over the years, but I'm starting to think my efforts aren't even useful. I imagine these companies have "shadow profiles" of my voice anyway. Should I give in and embrace Alexa, Siri, OK Google? |
While I'm somewhat concerned about the privacy implications, it has been very convenient. The biggest problem it has solved is "how to listen to music", and if that isn't a pain point for you, you don't really need it.
My kids have told Alexa to recognize their voices. I have several boys, and Alexa really can't tell them apart very well. (And, I haven't figured out how to edit the profiles; one of them taught Alexa he was John Cena, and I don't know how to correct it; another son uses a deep voice to tell Alexa he is Batman, which is kinda funny) I would not be concerned, at this point, about other people's devices keeping shadow profiles of you. (This statement probably needs to be reassessed every five years).
I did play with an open source voice assistant, but these are so much better, hardware- and software-wise. Would I take a self-hosted voice assistant if it reliably did most of the things I use Alexa for? Absolutely.
Personally I’m staunchly against it. Even Ring doorbell as well, they are rumored to begin using facial recognition and scan license plates. Too Orwellian for me.
My parents are surprisingly modern with technology, so they have Alexa in a few rooms, with most lights connected, and their garage.
Siri recently had a whistle blower contractor state they hear snippets of intimate conversations/sounds, with Siri not supposed to be activated.
These devices always have to listen, and for them to get better, the supervised learning must have a human say “yes/no” if the predicted output was correct for the input (in this case, activating on “Hey Siri”)
Some open source home automation systems are gaining traction, and once they’re a bit more mature, I’ll self host and begin my voice-assistance adoption. But until then, I’ll stick to the old ways :)
https://www.theguardian.com/technology/2019/jul/26/apple-con...
https://www.businessinsider.com/apple-whistleblower-siri-rec...
https://theintercept.com/2019/11/26/amazon-ring-home-securit...
IMO Alexa and Google are useless, half the time they don't hear you, half the time they hear you when you're not even talking to them and respond with something completely random. Google now refuses to talk to me because we messed up the personalisation and can't work out how to go back to just being anonymous.
Apple's Homepod however can't access the internet for answering questions but has a far far superior ability to hear you and respond.
We basically use our voice assistants to turn music on and off. Honestly that's all my friends ever seem to use these evil privacy invading devices for!
I trust Apple a bit more with my data than I do Google or Amazon.
There were lots of recording completely unrelated to anything I said to google. Like not just things said after a "Ok google" trigger. It had recordings at random moments.
Now its back in the box.
I had an Amazon Echo (Alexa) for some time. The only thing I used it for was toggling a smart bulb. Everything else was useless. Speech recognition isn’t reliable enough to be trusted for anything that impacts the real world like sending texts or ordering things and discoverability of what commands are supported is non-existent.
I'm not sure if you should embrace them outright, but I think it's useful to recognize there are some situations where they are more benefits than disbenefits.
I care about privacy, but for me, I'm willing to accept a certain tradeoff between privacy and utility. Voice assistants are genuinely useful for hands-free operations.
When I'm driving, I ask Siri to play specific playlists or podcasts, to turn the volume up, weather, navigate to this or that place. I take the issue of distracted driving very seriously so I try to never touch my phone while my vehicle is in motion. Siri is helpful in these situations.
At home, I only ask Alexa a couple of things: "weather", "set timer for 5 minutes" (for when I'm cooking and my hands are wet), "connect bluetooth", "play brown noise". I rarely deviate from these requests. But YMMV.
https://www.descript.com/lyrebird-ai
Alexa can already talk like Samuel L Jackson.
https://www.amazon.com/Samuel-L-Jackson-celebrity-voice/dp/B...
It's a risk, but to me it's also a part of living in this century. Human voice impersonators have always existed. When Photoshop became a phenomenon, everybody knew that photographs could be manipulated. SFX departments have always been able to produce fake scenes -- and now with deep learning, more people are able to do it (albeit somewhat badly). Not advocating that anyone feeds training data to voice assistants, but in the end it's a personal decision based on what you value.
For me, it's impossible to uninvent technology, so all we can do is to manage the risks.
I think I'm going to get rid of it.
Besides playing some music with voice control or telling me the news I have absolutely no use for it.
I can do those tasks with very few clicks and a Bluetooth speaker.
It was cheap enough to give it a shot but i wont be investing in the ecosystem.
I couldn't figure out where the hell the option to change what network it connected to was in the app and gave up.
On my phone Google only listens if I trigger it by swiping up on the home screen (maybe my faith is misplaced...), it's convenient for setting alarms and adding calendar events (the calendar GUI always scares me). Otherwise there's nothing always listening in my home. I have to unlock my phone and press a button to turn off some lights, but I can live with that; the alternative would be buying ZigBee light switches, not a voice assistant.
I think I'll do this. I don't feel like I can reasonably protect myself from data-snooping companies, whether I use their products or not.