Did some research on this a bit ago -- I believe the emerging CS term for this is "undocumented wake words." Phrases like "hey Google" are wake words (they activate your device's listening function), and the ones you know about are documented. Naturally, undocumented wake words would be ways of activating that function that you don't know about. I don't think there's a lot of literature on them, and no direct evidence, despite some papers on false activations, how much of a conversation is recorded when listening is deliberately activated, etc. This research was supposed to culminate in an experiment of my own on the topic, but my partner fucked it up, and I'm totally not mad and totally don't think Google got to him.
I could see a few ways undocumented wake words could work (actively recording literally all the time doesn't seem likely):
- It triggers on brand names.
- It triggers on topic keywords ("plumber," "bike," "dog").
- It doesn't trigger the listen/record/analyze function at all (like documented wake words). Instead, it counts how often keywords in one of the above categories are uttered and then sends the count back in lieu of a recording. This would be sloppier, but probably easier to design and hide, and if someone is constantly saying words you think cat owners use, you can guess they're a cat owner.
There are at least three other explanations for the phenomenon of getting an ad/video recommendation/etc. seemingly right after you speak about it:
- You're getting a hyper-targeted ad based on other things you do, not your conversations being recorded. See the famous example of Target sending an underage woman an ad for maternity items before she even knew she was pregnant. Had she told a friend she thought she was pregnant immediately before receiving that ad, she might have concluded her phone was listening.
- You searched for XYZ a few months ago (or emailed or texted about it), forgot, and now mistakenly believe you had only spoke about XYZ once and that must be how Google knows you're interested in it.
- Coincidence, maybe made more probable by some broader interest of yours. If you watch a few NFL highlight videos and your location data puts you in Seattle, you're probably going to get a Seahawks ad at some point, and it might be right after you talk about the Seahawks for the first time with a friend.
Minor nitpick but “underage woman” is a weird phrase. She’s a child, not a woman
I'd add the location of the user/device as a fourth explanation. If the feature is enabled, it can promote "relevant" suggestions.
Yeah. I've seen people say "it's because it uses the searches of people around you to target ads" but then I'll get something very specific, that I've only mentioned to like one other person, and I know they didn't search it.
One trick of ad tech companies is they try to guess the maximum amount information about you and associate it with your identifier & they don't care how accurate it is. If other people visited webpages that you also visited then they can use that info to infer additional attributes.
Some of my co-workers were discussing a show that I haven't watch. I never even looked up anything related to the show. A day later I had a suggestion to watch the season finale for it... I mean it could just be coincidence, but I honestly wouldn't be shocked if they were somehow always listening via phones and whatever other devices.
im pretty sure one thing they do is give you recommendations based on the people who are often in your geographic area. So if your coworkers watch the show and youre always near them, they'll give you recommendations based on that. When my sister comes home from college I always get Taylor Swift and Disney Channel type recommendations, then when she leaves they miraculously disappear
Ad tech companies use this technique less because advertisers are worried about ads that are too personal or creepy. Disney spends billions on ads and doesn't give af so ad tech companies will use whatever means necessary to pad their numbers.
I don't think they've got AI reading your screen, but they definitely have AI listening to conversations and noises that happen anywhere in the vicinity of a powered-on phone. I think that this, combined with the unfathomably powerful, dystopian network of data and analysis thereof is enough to explain all these little """coincidences""" we see with internet ads. And we only really notice when it hits the nail so directly on the head that you think, "wtf, how are you reading my mind???" But other times it's laughably off the mark, like when it shows me advertisements about the all new bucket of fresh uncooked animal guts from Local Restaurant, or shows me ads for reversing male pattern baldness, or for child care products, etc.
I don’t think they’ve got AI reading your screen, but they definitely have AI listening to conversations and noises that happen anywhere in the vicinity of a powered-on phone.
I think the other way around is more likely. It's well known that Google reads your Gmail emails (and their attachments), so odds are they read whatever you pull up in Chrome, or on your Chromebook, or your Android, etc. There's a good chance they can pull your data from non-Google sources, too, given how much shit has contacts with Google at some level. And if they aren't, Amazon, Microsoft, or Apple probably is.
On the other side, we don't have any concrete evidence of Google turning your phone into an always-on listening device (at least not on a mass scale). I'm guessing that would be a lot more computationally intense, too, although that's out of my depth.
I think it's due to the other users on the site then making halo-related searches, so you get adverts for it because there's reasonable likelihood that you'd be interested too.
That weird phenomena of talking about stuff then getting adverts, while I absolutely despise all advertising agencies and wouldn't be surprised if it is worse, I reckon is down to other people who were also exposed to whatever webpage it was that made you think/speak about it who then go on to search it in Google, and therefore your profile gets some likelihood associated with it that you'd be interested in that too and worth an advert. Analysis of all microphone data for every user is incredibly computationally expensive, and their targeting gets better every time somebody ignores/clicks on/spends screen time on an advert anyway so they wouldn't even need to collect and analyse all microphone etc. data to achieve the accuracy they do, and that's waaaaay more terrifying than "they listen in to our conversations". At least that way they'd be collecting accurate data at the source, and not being capable of reproducing it from the behaviour patterns of everyone else around us being exposed to the same things we are.
:screm-pretty:
That weird phenomena of talking about stuff then getting adverts, while I absolutely despise all advertising agencies and wouldn’t be surprised if it is worse, I reckon is down to other people who were also exposed to whatever webpage it was that made you think/speak about it who then go on to search it in Google, and therefore your profile gets some likelihood associated with it that you’d be interested in that too and worth an advert.
That doesn't account for the times that it's something that's entirely offline that makes you say it.
I tend towards this understanding, that these freaky coincidences to do with conversations or talking out loud are more a sign of how refined the actual tools that we know they use - the tracking of everything you do online and cross-referencing of that with everything millions of people vaguely similar do online - actually are.
I worked on this tech. Everyone collects your data as you browse the internet and sells it in the data market, even your ISP.
I read a Twitter thread by a compsci guy saying that he visited his mom’s house (connected to her wife etc) and in the coming days got ads for the brand of deodorant or toothpaste she used; that sort of thing.
His contention (and I think he had some industry experience to support his conclusion) was, as I remember it, that all sorts of companies have all sorts of data (like credit or loyalty cards associated with names, purchase data, etc). Data is widely sold, it just seems highly possible to pipe all these tables together .
I dunno. I guess the question is how cost effective something like this would be, or whether something could be more important than maximizing profit
"Customer 360" has been an ad tech buzzword for many years now. It means they know every facet of your existence so they can look at a useless analytics dashboard.
Literally any time I talk about take out food Uber eats will pop up with a notification that evening.
I was at the hardware store the other day trying to buy a palm sander and they had a really obscure brand I'd never heard of. I said it out loud and then decided to look it up to see if it's any good. I typed the very first letter of the brand name and google fully suggested the entire thing (3 words long)
