this post was submitted on 02 Sep 2024
156 points (99.4% liked)

technology

23289 readers
59 users here now

On the road to fully automated luxury gay space communism.

Spreading Linux propaganda since 2020

Rules:

founded 4 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] Camdat@hexbear.net 50 points 2 months ago* (last edited 2 months ago) (25 children)

This is maybe my biggest pet peeve. These companies are not listening to you in any meaningful way.

You can trivially confirm this by hooking up your home network to Wireshark and filtering packets.

Other reasons:

  1. They can get all of this information elsewhere: searches, ad pixels, location capturing etc.
  2. Processing audio data is basically impossible on-device in a useful way, and the network infrastructure to support mass transcriptions on the cloud would be on the order of billions.
  3. It would be a massive endeavor to cover up the millions of hours of audio data that would need to be analyzed by the lowest paid and most unhappy workers in the industry (content labelers and moderators)

Now I'm sure this is some marketers wet dream, but the logistical and PR nightmare this would create dissuades all but the dumbest ad agencies. This is mostly just terrible tech journalism.

[–] blame@hexbear.net 37 points 2 months ago (7 children)

Not that I disagree with your conclusion because there's an even simpler way to check if an app is listening: iOS and Android will tell you the mic is being used... Anyway, we do have always-on NNs listening for keywords ("Siri,", "Hey google", "Alexa") so I agree that full ass voice transcription like whisper will run like dogshit on your phone they can certainly run a much much lighter model to pick up a handful of keywords.

[–] Camdat@hexbear.net 13 points 2 months ago (1 children)

Sure this is definitely true. I should clarify that single-word NNs do run on-device all the time, but those require specialized models that are trained only on those keywords. Once those models trigger they need to send everything else to the cloud.

[–] blame@hexbear.net 15 points 2 months ago* (last edited 2 months ago)

I agree. If I was going to do something like this for advertising though I wouldn't really care too much about what people were saying so instead I'd just listen for some limited set of keywords (maybe for some of my top paying advertisers) and serve ads for keywords that hit recently. Keep it all on device until an ad actually needs to be served.

load more comments (5 replies)
load more comments (22 replies)