We've all (or know of someone that) had that experience of mentioning something out loud and some time later having an add for that appearing somewhere on the web.
I know that there is a fair share of confirmation bias in this stories, and that network connections + interests can be a quite powerfull way of working those add engines.
Being a data scientist I know how Criteo, Google ads and Facebooks work, by having worked with them. And I also have a faint idea of how they may work under the hood.
However, I recently had a discussion with a couple of friends where the only way they could have gotten a particular add is if voice triggered adds were a thing.
I believe that it would cost so much money, given the conversion rate (even if abnormally higher for a voice enabled add), computing costs (on Google Cloud or Aws), engineering costs, amount of audio to be processed for speech to text that it would not be feasable.
Also, the amount of adds I can come up given a basic list of connections + interests seems way much bigger then given Gbs of audio to process and come up with adds from that.
My friends argue that, given what companies can afford, it would be peanuts for a share of their budget to be used in such applications. Also, they argue that basically any app those days have acces to microphone, and I am not sure how those permissions are handled by the OS (IOS or Android).
So are companies hearing us all the time? Are there any add companies that claim to do that?
I know that there is a fair share of confirmation bias in this stories, and that network connections + interests can be a quite powerfull way of working those add engines.
Being a data scientist I know how Criteo, Google ads and Facebooks work, by having worked with them. And I also have a faint idea of how they may work under the hood.
However, I recently had a discussion with a couple of friends where the only way they could have gotten a particular add is if voice triggered adds were a thing. I believe that it would cost so much money, given the conversion rate (even if abnormally higher for a voice enabled add), computing costs (on Google Cloud or Aws), engineering costs, amount of audio to be processed for speech to text that it would not be feasable. Also, the amount of adds I can come up given a basic list of connections + interests seems way much bigger then given Gbs of audio to process and come up with adds from that.
My friends argue that, given what companies can afford, it would be peanuts for a share of their budget to be used in such applications. Also, they argue that basically any app those days have acces to microphone, and I am not sure how those permissions are handled by the OS (IOS or Android).
So are companies hearing us all the time? Are there any add companies that claim to do that?