You may be thinking about this article about how Target knew that a woman was pregnant before her family knew: https://news.ycombinator.com/item?id=3598558

I wish we had an update on what the situation looks like today.

Yeah AI is taking a lot of damage when the actual problem is capitalism

More perfect and consumer reports did a joint investigation and have uncovered this already happening.

https://youtube.com/watch?v=osxr7xSxsGo&pp=0gcJCR4Bo7VqN5tD

>Maybe I am getting paranoid. But to me a lot of innovation in the last years feels openly hostile and primarily designed to extract maximum money while providing only little actual benefit. AI will just accelerate this trend.

This is the natural evolution of capitalism in a system where regulation is weak. Political donations from companies need to be outlawed and the proper regulatory environment restored. Capitalism is like a powerful engine that will happily rev to redline and tear itself apart if the engine lacks a governor. Regulation is that governor.

"It annoys me that big-tech marketing has made most people believe that "personalised advertising" means they get ads which are more "useful" to them."

"relevant" is another term seen in addition to "useful"

But "relevant" is relative

For example, "relevant" to what?

It's only if Big Tech has collected data about the ad target and, e.g., made some guess about their intent, that the ads could be "relevant"

Whether the ads are truly "relevant" is a question for the reader. The term "relevant" might just be marketing fluff

Either way, Big Tech will keep the data vacuum humming

Where the AI makes a difference here isn't regular personalized advertisement (which already isn't all that great, based on the percentage of ads I get for products I would never consider at all, or are downright offensive to me), but in understanding your existing consumers, and attempting to do habituation effects.

So imagine you have a bunch of money, watch sports while drinking and are bad at math, and therefore are considered to be a great target for sports betting companies. Making sure you get used to betting most of the time you watch a game is very valuable for the company, so just realizing what teams you like, when they play, and what kind of bets might look good to you, but are really pretty iffy is very valuable to them. Just like they would love to know when you are bored, or depressed, and maybe betting on the game that is going on right now would be appealing: A level of access to you that, say, a casino, or a bar that you haven't visited in a while just doesn't have. And habituation models are simple, you don't need a very expensive system to know when offering you a discount to entice you to don't break a gambling streak will pay off

Now that is using AI in ways that are quite antisocial by most standards: the current advertisement that tries to sell me hair growth when I have all my hair isn't all that scary.

This is one of those places where it's worth disentangling the status quo from an optimal alternative.

Currently, every factlet you leak to one of these systems poisons them toward their profit (and almost unanimously against your best interests). Advertising draws your attention away from the products that would make your life better (cheaper, heathier, tastier, whatever) and toward profitable alternatives.

It doesn't have to be that way though. You physically don't have time to research every thing that exists, or even to hear about every possible product in passing. Supposing some of those would improve your life on average, is word-of-mouth really the most efficient way we can come up with to tell you about the things you do actually want to spend your money on? In theory, this is a great business -- customers want to spend money, companies want to sell things, and the information/discoverability asymmetry means that companies are inclined to get word of their products out there with customers _also_ wanting to hear about those products (if they're sufficiently personalized). If "advertising" were good enough, I'd pay money for it.

That only falls apart because of a lack of trust and ethical behavior. Instead of being treated like the information market it is, it's thrust onto individuals to try to prey on their weaknesses.

if you draw a venn diagram of all the stuff i get advertised on and all the stuff I actually buy, the two circles are in completely different locations with virtually no overlap whatsoever. the only time i get ads that are even remotely related to my purchases, are only ads that come after I've made the purchase and am done. personally, i don't see how they make any profit off me whatsoever.

i think the whole "personalised advertising" thing is way oversold and more for the benefit of a sales pitch for the advertisers but reality is far from it. google makes their money on volume, not accuracy. and so all the "information" they collect, doesn't seem to translate into more targetted advertisement.

It annoys me that there aren't laws to prevent this. Or that anti-monopoly law wasn't effectively used to separate the largest advertising company in the world from a consumer software browser product which is clearly being used to facilitate and amplify these outcomes.

I'm thoroughly annoyed that adblockers aren't installed by default and require an opt out to disable. This will not at all touch first party advertising, but, it will put a huge dent into dynamic third party advertising. Which seems to be the source of the problem you describe.

Our government is genuinely failing to represent the majority on this issue.

To be fair, if I choose to buy something, it's almost by definition because I consider the thing useful. It's pretty rare that I purchase something I'd learned about from an ad, but I have done so a few times and benefited from doing so. How is anyone else to determine whether I've been adversely manipulated, i.e. whether the cost of the thing outweighs its benefit to me?

Buying something from an ad isn't really fundamentally different from being influenced by an HN post. For example, thanks to https://news.ycombinator.com/item?id=46294574, I read up on and have decided to experiment with TLA+ on my next project. It's really no concern of mine what Martin Kleppman's commercial interests may ultimately have been in publishing that blog post; I received value from the information all the same.

Personally, I'm not particularly bothered by ads per se. I'd be more bothered by information being withheld from me during searches, e.g. if Brand A could pay Amazon to delist Brands B and C from organic search results, since that would directly guide me toward less optimal purchases. But as far as simply going about my day and seeing a billboard or promoted social media post every now and then, I don't see the big deal. It's generally easy to ignore, it costs me almost nothing, it occasionally helps me, and ultimately it funds a lot of things I like and take for granted (e.g. Chromium, Firefox, and Android).

I'm not saying that people who routinely waste money on irrational purchases don't exist. I just don't find that to be a compelling argument against the existence of a particular market which overwhelmingly benefits almost all of us.

I do have quite strong concerns regarding aggressive data collection, however, and I certainly wouldn't opt in to greater erosion of my privacy — but I see that as a separate issue. To the extent that ad-driven revenue models provide an incentive for companies to facilitate greater privacy invasion, I agree that it's a significant concern which warrants much stronger pushback from the public than it receives. I just think it's important to highlight that mass data harvesting per se is the major issue, more so than any perceived manipulativeness of the fact that brands pay money for exposure.

Then of course there's the issue raised in this post, which is yet again another matter entirely. I'm all for using AI to optimize pricing and efficiency, but "dynamic pricing" as described in the article sounds like a euphemism for price discrimination, and should be more strictly regulated IMO regardless of whether or not AI is involved.