No, the text ad which is targeted (crudely) using tons of data and some statistical methods.

It ain't gonna pass the Turing test anytime soon, and it's not more of an "AI" product any other product with basic algorithmic based matching is.

Google are a company whose core organizing principle is using statistical methods over obscene amounts of data. Apple is a company whose core organizing principle is selling desirable consumer products with class-leading design through the single most impressive physical supply chain in the world.

OP has a point.

>whose core organizing principle is using statistical methods over obscene amounts of data.

Sure. And we didn't use to call them "AI companies".

When someone says "Google is an AI company" they're obviously not talking about AGI but about applied stuff like ML and IR.

Well, then how about "Google is an ML company"?

Though I fail to see any application of machine learning to write home about in their core business -- ads.

Search yes, they do a good job on that.

But given the amount the data they have on us (browsing history, profile, mails in Gmail) has anybody been that impressed with any targeted ad? I've seen targeted ads from "dummy" networks like the Deck that are 1000% times better for my interests.

> Well, then how about "Google is an ML company"?

Sure, that's fine, but since ML is generally considered to fall under the AI umbrella your statement is more precise but not more correct.

> Though I fail to see any application of machine learning to write home about in their core business -- ads.

I know basically nothing about our ads (I work for Alphabet), but I believe there is some kind of ML involved.

When I buy Google ads I tell Google what search keywords I want my ads to match.

Then you're wasting money.

I'm sure this is pretty proprietary information, but I do wonder what the value add of the AI-targeting is. Is it a relatively small bonus on top of a solid platform and audience for ads? Or is it the killer feature?

Recently I searched Amazon to read reviews of a book I bought on Kindle a few years ago. Now every website with Amazon affiliate advertising is showing me ads for that book.

Considering that Amazon could have run a simple account search to determine that I am definitely not interested in seeing ads for a product that I already own, I think that the AI-targeting ad revolution is a while off yet.

That is unrelated to Google, but I haven't seen anything there that would suggest they are running anything smarter.

I have the same experience. On my machine, only Safari is "vanilla" - other browsers have various levels of ad/js blocking. So the only time I see ads is in Safari. I can't believe they're still feeding me Newegg ads for components that I already own and have bought from Newegg. I understand that's not as trivial a problem as it sounds, but I can't help but think companies are overpaying for ads.

i have friends who work for retargeting ad-tech companies. That's those firms that bid for and deliver ads for things you just maybe bought.

The deal is those ads show up when the ad space is the cheapest. It's known as remant inventory. So the retargeters can pick it up for very cheap.

The deal is the companies KNOW that some people see ads for things they already bought. But some people are seeing ads for things like only considered, and bombarding them with ads does affect behavior. The question is, is the increase in revenue worth the cost of retargeting.

The answer is very much yes. Even as it annoys people who already bought it, which doesn't even factor since they already made the revenue from your purchase.

To be fair, Google is in the business of making money on ads. If the customer wants to throw money on a stupid ad buy, then that's not particularly their concern.

Yep, Google does this exact same thing. I search, I buy, and only then I see ads for a product I already own.

It's just as likely that they have done extensive testing and found out that for whatever reason, ads for products you already own still encourage you to buy other stuff.

A lot of assumptions we make don't hold up when you start gathering real data.