Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
AI Microsoft Google Privacy Technology

41% of Voice Assistant Users Have Concerns About Trust and Privacy, Report Finds (techcrunch.com) 47

Forty-one percent of voice assistant users are concerned about trust, privacy and passive listening, according to a new report from Microsoft focused on consumer adoption of voice and digital assistants. From a report: And perhaps people should be concerned -- all the major voice assistants, including those from Google, Amazon, Apple and Samsung, as well as Microsoft, employ humans who review the voice data collected from end users. [...] While some users may not have realized the extent of human involvement on Alexa's backend, Microsoft's study indicates an overall wariness around the potential for privacy violations and abuse of trust that could occur on these digital assistant platforms. For example, 52 percent of those surveyed by Microsoft said they worried their personal information or data was not secure, and 24 percent said they don't know how it's being used. Thirty-six percent said they didn't even want their personal information or data to be used at all.
This discussion has been archived. No new comments can be posted.

41% of Voice Assistant Users Have Concerns About Trust and Privacy, Report Finds

Comments Filter:
  • Wrong headline. (Score:5, Insightful)

    by Gravis Zero ( 934156 ) on Thursday April 25, 2019 @10:46AM (#58489126)

    59% of Voice Assistant Users Are Too Trusting, Report Finds

    FTFY

    • by Anonymous Coward

      60% of people are complete idiots and would allow you to spy on them 24/7 in exchange for small conveniences.

      "Alexa, hold my beer." (Pun: she's a coaster.)

      • by jrumney ( 197329 )
        That is the 41%. The other 59% are ignorant of the spying, which is a different type of idiot.
    • by godrik ( 1287354 )

      Actually I think it is more:
      59% of Voice Assistant Users Know They Are Spied On.

  • 99% of people are more worried about these devices than the real eavesdropper which belongs to the smart phone.
  • by dryriver ( 1010635 ) on Thursday April 25, 2019 @11:04AM (#58489236)
    If a desktop computer of that era could handle voice recognition of rather complex spoken texts, why the fuck do smartphones, smartspeakers and such need a _backend_ tethered into the internet to understand what you ask the device to do? Voice recognition does NOT need TFLOPS of processing power to work. A smartphone can certainly do it, and a smartspeaker - which I would never bring into my home - could do it with a relatively cheap SOC. So why this listenallthetime -> sendtoserverfarm -> voicetotext -> parse -> sendresponseback dance that is being done? These are corporate snooping devices, that's why. Imagine JUST HOW MUCH DATA an always-on smartspeaker setup that listens to half your house can gather about you. Maybe you have conversation about bills, loans and taxes? Maybe you talk about food or football all the time? Maybe you say "I'm real excited about the upcoming Star Wars The Rise Of Skywalker" to your roommate or spouse. These fucking devices appear custom built to gather data, data and more data and phone home. Why else would voice-recognition - which ran on bog ordinary PCs 15 years ago - need an elaborate server backend??? You could just say "Where can I find Feta cheese?" and the device could - in an anonimized fashion - google "Feta cheese + geolocation". Voila. Done. Smartspeakers are for Dumbconsumers apparently.
    • Because unlike Dragon dictate it works?

      Speaker independent and without hours of training that is...

    • by godrik ( 1287354 )

      Well, that's a good question. I have a few ideas:

      1/ It makes it cheaper. Right now, you need virtually no processing power in that device, just a network chip, and enough processing power to drive it. You can push all the processing in the cloud. If you save $3 per device and you sell millions, that could make a difference on the bottom line.

      2/ It is easier to upgrade. If you change voice recognition algorithm and the new one is more computationally expensive, you don't need to change hardware in millions o

    • by ljw1004 ( 764174 ) on Thursday April 25, 2019 @12:32PM (#58489750)

      I bought Dragon Dictate back in the 1990s. If a desktop computer of that era could handle voice recognition of rather complex spoken texts, why the fuck do smartphones, smartspeakers and such need a _backend_ tethered into the internet to understand what you ask the device to do? Voice recognition does NOT need TFLOPS of processing power to work. A smartphone can certainly do it.

      I too used Dragon Dictate. It was terrible. I spent weeks training it, and it still got mediocre results. My brother got progressively irritated with training it and his annoyance crept into his voice, and he ended up with a Dragon Dictate that only understood him for dictation when he was speaking in the same cross voice.

      I agree that it doesn't need TFLOPs of processing power to work. That misses the point. What it needs is TRAINING - huge, huge training sets, billions of conversations in all kinds of scenarios. No individual is going to waste weeks training their own device only to discover that their training was incomplete anyway. These voice assistants rely crucially on being trained from data from all of their customers in all kinds of different situations.

      In twenty years once enough training data has been amassed, and voice recognition is halfway decent, then companies won't need online hooks. Until then, whichever company avoids online integration will fall behind in the race to get decent voice recognition.

    • why the fuck do smartphones, smartspeakers and such need a _backend_ tethered into the internet to understand what you ask the device to do?

      Because Dragon Dictate doesn't house the information dispersed over the Internet that products like Alexa look up to answer the questions you ask it.

      Neither do smartphones or speakers. You need the Internet to do that.

      That doesn't dismiss the fact that Amazon, Google, et al do bad(tm) things with your data and don't respectfully disregard or anonymize it before storage.

      • Also, it's the "AI" involved. It's constantly getting better due to users' data. I'm not opposed to making a product smarter with the data I produce by interacting with it but it needs to be anonymized, transparently.

  • These devices used to be known as BUGS. You would for instance sneak one in to an embassy and listen in.

    Now people think they are just devices to order stuff.

    But they ARE still surveillance devices that are used to spy on the room they are in no matter what the TOS says.

    • The key feature of a bug is that you don't know it's there and you don't know they are listening.

      So as far as I'm concerned, MY smart-speaker and MY phone are not bugs in that sense of the word.

      YOUR smart-speaker and YOUR phone on the other hand....

  • And Yet... (Score:5, Insightful)

    by WankerWeasel ( 875277 ) on Thursday April 25, 2019 @11:14AM (#58489306)
    And yet that hasn't stopped them from continuing to use them. "I'm concerned about the effects these cigarettes are having on my health but darned if I'm gonna stop smoking them."
    • by jwymanm ( 627857 )
      Because it is a non issue. That is until the government gets involved and demands a backdoor to that "horrible data collecting we aren't doing yet!".
  • by Cajun Hell ( 725246 ) on Thursday April 25, 2019 @11:44AM (#58489438) Homepage Journal
    I really would expect it to be somewhere around 0%. If you care about privacy, doesn't that really mean you aren't using one of these external "assistants?" (Precisly because you know that it compromises your privacy. Why would you be sending internal communications outside?)
  • 59% of Voice Assist users don't give a shit about their privacy.

    But it gets worse. 41% of Voice Assist users know that these things spy on them and STILL use them.

    • Don't forget those little ToS changes where they'll keep everything that you say for years to "train their algorithms." Tomorrow Google may change their ToS and say "fuck it, we'll keep everything you say and you can't change that." I doubt half of the idiots using their products or Google Assistant will even bat an eyelash.

  • Sick as it is, 99.9% of them will continue to use it anyway.

  • by ET3D ( 1169851 ) on Thursday April 25, 2019 @01:40PM (#58490372)

    This survey is based on results from 5000 Suzy users. Even after googling, I have no idea who Suzy users are, whether anyone can join, and if not, how they are chosen, how they are asked questions, and whether they can choose which questions to answer.

    Without such knowledge, these 'statistics' are meaningless, and depending on the answers, they may be completely wrong (for example, if users can choose whether to answer or not, which would make the samples self-selecting rather than random).

  • Unfortunately, I don't have any statistics, or p-values to share, but I swear YouTube ads are targetted based on keywords that are spoken within range of my Google Home Mini. I received my Google Home Mini free with an ebay promotional purchase, so I was skeptical from the start. However, my roommate started half-seriously joking with me that it was silently sending feedback based on keyword triggers. To test it, we started very obviously saying "Pizza Hut" in various contexts. Within 48 hours, I saw an
  • Once upon a time, I was charged (it is debatable if that was a promotion or punishment) with evaluation of "smart" solutions developed by a sister company. As a habit, even after escaping that addition duty, I tried most new "solutions" in the market. I am still to find a device or program that performs a function that becomes easier by giving commands talking instead of typing, touching a screen or pushing a button.

    Generally speaking, besides security risks, false recognitions after hours of training is

    • I am still to find a device or program that performs a function that becomes easier by giving commands talking instead of typing, touching a screen or pushing a button.

      "Alexa, play KFBK." That took me about five seconds to say, and I can say it from anywhere in the room.

      What button do I press to have KFBK start playing? What do I type on to do that? My desktop computer? What exactly do I type? If it takes longer than five seconds and requires me to turn the computer on and/or go to my desk and plays on my desktop speakers, then it is not easier.

      Yes, IF there was a small box with a button on top already programmed with "play KFBK" then that might be easier, but I don't

HELP!!!! I'm being held prisoner in /usr/games/lib!

Working...