This is a good summary
View: https://www.youtube.com/watch?v=2wjy-v7RXtg&t=22s
tl;dr; it's still not happening.
"the phone", ie the actual operating system, is not listening to you and serving you ads because of this. This has never been proven, and people are just really really bad at understanding technology, or confirmation biases.
If you install shitty scam apps that has ad systems in it that listen to you while they're open, that is on you and your bad app choices.
Now I have to watch this video too, because basically until now I have always thought "yes it is totally possible, but why would they be doing this, blatantly lie about it, get away with it, when they can get just as good info most of the time from proxy data in your pocket instead of having to listen to you all the time for advertising fodder.
My take on this has recently evolved and been:
- Could it be happening? Yes, but I thought, not likely, until recently, for mostly this reason:
-
Listening to devices 24/7 was costly noise for a company to parse.
- But if you think about it, if they could crack the solution to this problem, i.e. "How do we streamline organizing and acting upon information we hear 24/7 to ignore unimportant information but also document information we do value to sell our products" it's the golden goose for advertising $$$.
- Until the last couple of years, this seemed like the golden egg that had yet to be cracked. But then you got articles recently explaining that Amazon lied about their "smart tech" figuring out how you shop at Amazon Fresh locations....instead they just mass hired cheap labor in South and Southeast Asia to watch you on camera and manually document what you bought. Not only is that deceptive, it's still extremely costly at-scale, but they minimized cost where possible: hired cheap labor overseas and used their big wallets to sustain that initiative. while deceiving shareholders that they had some kind of edge on sophisticated technology that allowed them to do this mostly unsupervised. This demonstrates that companies not only have the means, but the will, to deceive us and actually brute force their way through data analytics...but now, things are even more precarious because:
- LLM AI has gone mainstream and consumer-grade, meaning they now have a sophisticated tool that can "brute force" through the data noise with minimal human oversight. Instead of having to log all info it hears, AI can, to a degree of accuracy, determine which key words and phrases out of all the audio it captures are worth logging or selling advertising about. I surmised that this was now not only possible, but combined with other articles, we can see companies now have the means, the wallets, the intent, and the deception behind, using audio from your devices, to sell your life to others, or even report to authorities (which I guarantee will disproportionately affect minorities).
- It's especially eyebrow-raising to read this article and the extents to which this is going. I knew I wasn't crazy when I had some private convos in the last few years where I literally unplugged all my smart home devices and put phones in another room to talk. Part of me thought that was insane, like do I really live in the reality where companies are doing this, but there have been NUMEROUS suspicious examples in my life in the past couple of years where there was simply no other explanation than either "very good guessing" from proxy data like proximity, contact lists, GPS, etc. to guess what you're talking about, OR these companies have now engaged tech that lets them lazily listen to you 24/7 for what they want to hear. I have had instances where me and 3 friends were talking about something we've never searched for on our devices, and at least 2 of us suddenly got search results in our online shopping that suggested the very brand and thing we were talking about, over an hour after the conversation was over and we had transitioned ideas/activities/topics. It was so creepy that I never forget moments like that.
This is very concerning to me as an AI data scientist. Just because the tech is good and powerful at certain things (like pretty much it is invaluable to science and engineering innovation), when you make tools like this consumer-grade and available to pretty much anyone for almost free, with little/no global or national ethical oversight, you risk huuuge human rights violations and ethical trespasses that I'm not sure how to fix.
The genie is out of the bottle.
futurism.com
