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?
21 comments
Depends on your own personal convictions. Once they have your data, they have it (unless laws come about forcing them to delete it, which is unlikely)
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...
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...
I'm also very against it, but I'm not convinced they don't already have whatever data they'd gain from using the assistant.
Regardless of privacy issues the functionality they provide is very limited and not reliable in my opinion.
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 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.
Really? The recognition has worked pretty well for me and people I know.
I tried a few (siri, google home, echo) and I don't think they are any more convenient than typing so they probably aren't worth the privacy loss. If you don't care and can find some use case for it, go for it.
The use case would let me be a lot more handsfree.
I thought there was an opensource version of this you could load on a raspberry pi? No?
Do you want to? It depends only on your decision. Are you being peer-preassured? Would having a voice-assistant change your life for the better?
It would. I would learn to use it for everything I could. Setting timers with voice will be a big one for me. Changing music, hanging calling/hanging up the phone.
There was a really good article on 'locking down' voice assistants I read recently in Make: (Spring 20) where the author proposes 3d printing 'hats' that sit over your Alexa / Google Assistant and has electronics inside that listen to you and pass your message on to the voice assistant. Their approach was to stop the device from listening to you all the time.
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.
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.
I'm giving the Google home mini a shot right now.
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 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 had a google home mini for a while and made limited use of it till I moved.
I couldn't figure out where the hell the option to change what network it connected to was in the app and gave up.
I couldn't figure out where the hell the option to change what network it connected to was in the app and gave up.
My MIL gave us a Google home mini as a gift. I plugged it in and played with it for a few days in hopes to embrace it (after not using it for like 3-4 weeks). Then one day I found out I can see what recording were being made somewhere inside the google accounts page.
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.
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.
If you are looking for a privacy-respecting and open source voice assistant, check out mycroft.ai: https://mycroft.ai/, https://hn.algolia.com/?q=mycroft
This is HN. So you're likely to get a bunch of "no"s.
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.
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.
I understand the utility, I had a friend with a HomePod who used Siri for everything and I became jealous of how smoothly he used it to his benefit. I just don't see how in the future my voice is going to be protected. If I don't use it, they'll probably still have my voice.
Re: voice -- yep, you and every celebrity -- it's already possible to synthesize voices from a small sample.
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.
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 don't get the question. If you're resisting, who's forcing you?
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.
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.
We have an some Echo Dots.
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.
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.