67 comments

[ 0.29 ms ] story [ 8.6 ms ] thread
What a garden path of a sentence!

I’m not surprised by this, unfortunately. One can hope that it does not materially affect (bias) output.

If it doesn't bias output how much would they pay for the ads? It seems like ads that also bias the output would be way more valuable.
When ads are inevitably folded into the prompts to the AI, I imagine it'll be fun for a bit to prompt the AI to react with disgust to the idea of having to serve some ad. Until they fix that hole.

I hate that the progression is so predictable. Yeah, clearly labelled ads today. Less clearly labelled tomorrow. Then they'll be indistinguishable from the text of the AI but still separate. Then they'll start folding it in to the AI prompts. Then they'll start folding it in at the training phase so the weights themselves drive you towards Thinking Ford First, this step being inevitable because it almost certainly will be by far the most effective of all of these techniques. Opus 7.2 Brought To You By Mountain Dew. Just a coincidence that every time you ask it to build an authorization system it writes code to ensure that the user has drunk a verification can, and is just incompetent at removing that code for some reason.

Sigh.

Enjoy this era of AI, folks, while the AIs are still developing too quickly for enshittification to catch up properly and the market and indeed the National Interest won't let anyone slow down so it can. But it will. It is inevitable. Make sure to at least keep a toe dipped in the open weights AIs, and any future truly open source AIs that develop.

Right now it’s only for the free tier, but given the recent trends I imagine it will make its way to paid tiers.

Just consider how even paid streaming services are bringing in ads, or how TVs were caught using ACR to spy on us.

Paid tier just means you're a more lucrative ad target.
Well they are being generous giving you extra 15 min video streaming which cost a lot to produce for free in a 45min show.

Soon, one need to be tipping advertising producing partners for their hard work.

nah this will never happen. do you really think they will do post training based on who pays them more?
No, they'll do a live auction from multiple models during reply generation. Google has the tech already. They are an advertisement company, you know: https://research.google/blog/mechanism-design-for-large-lang...

In short:

We investigate the design of auction mechanisms for aggregating the output of multiple self-interested LLMs into one joint output. We argue that this comes with a number of unique challenges, and we propose a simple token auction mechanism and illustrate our design through demonstration with an off-the-shelf LLM.

This era reminded me of late 90s web culture, and it's heading the same way.
> Then they'll be indistinguishable from the text of the AI but still separate. Then they'll start folding it in to the AI prompts. Then they'll start folding it in at the training phase so the weights themselves drive you towards Thinking Ford First, this step being inevitable because it almost certainly will be by far the most effective of all of these techniques. Opus 7.2 Brought To You By Mountain Dew. Just a coincidence that every time you ask it to build an authorization system it writes code to ensure that the user has drunk a verification can, and is just incompetent at removing that code for some reason.

I know close to nothing about the US regulation in this space but at least in Europe, when you pay for advertisement, it has to be always clearly communicated to the user. Are you assuming that they will just ignore the law in such a blatant way?

I would guess they’ll spend ages arguing about what an “ad” is.

Eg if I ask Llama 3 to find me the best headphones, I think we can agree its response isn’t an ad.

If Claude responds with a banner that Sennheiser designed and paid Claude to show, that’s an ad.

They’ll live in the probabilistic grey area between those. If Sennheiser pays Anthropic to add “Sennheiser makes really good headphones” to the system prompt, is it an ad if Claude recommends Sennheiser headphones? What if they pay to ensure their site is included in the web search tool results for any queries about headphones?

It’s sort of akin to paying Google to rank higher in searches if that were a thing they did.

I do think they should have to declare those, but they’re not traditional advertising to me.

> If Sennheiser pays Anthropic to add “Sennheiser makes really good headphones” to the system prompt,

I think you mean "OpenAI is partnering with Sennheiser to improve its AI."

Load up on verification cans because the companies who got away with flaunting EU advertising/tracking laws will get away with obnoxious ad-driven AI too.

(The cookie consent banners that you see everywhere were anticipated and forbidden in the laws themselves, companies just gambled that the EU wouldn't be able to enforce the law, and they were correct. I expect something similar to happen with for-profit AI steering.)

So, while it's tempting to be cynical and sarcastic, my honest answer is that I expect them to try it, and it would actually be a value judgment as to whether that constitutes ignoring the law.

Common Law-based systems prioritize the consistency of precedents. As a result, the real law is often built on the margins, to borrow an economics terms. Which means that in a sense, we find out what the law "really says" by someone first breaking it and someone else challenging them. What I expect to happen is that we will eventually get to the point that someone is going to judge that the value of trying this is more than the expected value of the resulting lawsuit, and then this will eventually get to trial, and then, a couple of years later, we'll "know" whether or not this is breaking the law.

But there is a sense, a real, non-sarcastic sense, in which until that point, we don't really know if it's illegal or not.

Most of Europe does not operate under Common Law, so this may seem strange.

Won't this be a bit of a problem when your new microservice includes an ad for Diet Coke? Has someone patented a mechanism for putting ad delivery into someone else's vibe coded consumer project yet?
I can't wait to need to drink my Mountain Dew verification cans to keep the slop hose running at work and keep paying on mortgage.
>while the AIs are still developing too quickly for enshittification

LLM's were born enshittified, there was no enjoyable era. It was made by the enshittification machine for the enshittification machine.

Marketing companies are already selling their ability to influence the training phase. I've seen entirely fake companies with fake products getting recommended for real queries, and the companies behind them selling this as proof to real companies. How would the models know the difference between a completely astroturfed e-commerce site and a real one? What data are they going to use when you ask for the "best natural deodorant" other than their corpus of scraped sites, more of which are fake everyday?
Who better to detect slop than the slop engines themselves?
Forget Amazon, ultimately this is going to dramatically alter consumer choice on a manufacturer level and OEMs won't be able to "SEO" (or whatever the equivalent ends up being) their way around it.

Users might not even notice - Will they prompt around it e.g., "Didn't Dyson used to make a hair dryer that was well reviewed?" if it isn't even recommended to them?

I regret to inform you that the equivalent has been around for some time already. They're calling it "GEO," for Generative Engine Optimization.
It looks like the old saying is true. The smartest minds in all of tech are working overtime to get you to click on advertisements.
The smartest minds in all of tech worked overtime to finally give computers the ability to lie to us. Excelsior!
It has ever been the case that garbage in, garbage out. AI is nothing new in that regard.
maybe the smartest minds in all of tech can’t do better.
The smartest minds in advertising convinced everyone they were in tech.
This is not meant to be tongue in cheek comment but has anybody figured out how the companies manage the "everyone is going to be replaced in a year" rhetoric while also working hard on ads that are clearly designed for humans to consume?
everyone is laboring under the assumption that things will continue approximately as normal because what the hell else do you do? how, as amazon, would you plan for the collapse of all economies and markets? sure, as maybe an amazon exec you plan for you and yours, but amazon the behemoth? its got nothing.
Even if they’re on a path to make economy obsolete, they still need money right now to get there.
Messages are tailored for the audience. The "AI will replace all jobs" narrative is used to sell AI to business leaders.
And who will these businesses have for customers when nobody else has a job?
It is the job of management to figure out how to make gains over the next 90 days, not make gains sustainable. The money available for consumers to spend in wake of their unemployment is the problem for whenever that issue occurs.

If there isn't a good way, you break the company up, sell it for scrap, give the resulting money to the shareholders, and they do whatever with it. It's the Jack Welch model.

> It is the job of management to figure out how to make gains over the next 90 days, not make gains sustainable.

This is the cancer that has killed so many otherwise good businesses here.

Oh, no question. But realize, that is their mindset.
the MBA logic based on a very biased and small sample size from a very select timeframe in the 1980s is that advertising during a recession leads to higher payouts when the economy recovers: https://knowledge.wharton.upenn.edu/podcast/knowledge-at-wha...

there's also a lot of language on essentially preying on people's insecurities by selling them an artificial sense of agency which is a fun reminder of how manipulative marketers are

I believe "classical" advertisements, as in those intended to convince the average person to purchase a tangible product, are on their way out.

I'm sure I'm not the only one who's noticed an increasing percentage of ads that fall outside of that category - either aimed at business owners and wannabe entrepreneurs, trying to sell some sort of get rich quick scheme, or just straight up scams.

Your mistake is parsing rhetoric as anything other than made up.
The amount of logic and thoughtful reasoning in your comment is so strong, its making my head hurt. Why are you raining on our IPO parade?
They want agent driven shopping and ads to sell to those agents
By making decisions under the assumption that society will continue.

Even under the reduction to absurdity of everyone's jobs being replaced in a year, there's still some system of allocating resources and they would still like more resources allocated to them. That system still likely bears resemblance to the current economy and convincing people/selling the ability to convince people is likely still a valuable ability to have.

> By making decisions under the assumption that society will continue

That seems completely at odds with "everyone will be replaced by AI"

> Even under the reduction to absurdity of everyone's jobs being replaced in a year, there's still some system of allocating resources and they would still like more resources allocated to them

I pretty strongly disagree. If everyone's jobs are replaced, I think society is over. I just don't believe that the UBI style future is going to play out in any kind of peaceful way

Companies are driven by people oriented at the next quarter's goals to maximize their stock portfolio's value. Nobody cares what will happen, everybody just creates narratives.
Did anybody think this wasn't going to happen?

They weren't going to burn a couple trillion dollars of capital and lose money on every API call just so you could ask it to tell you why your cat brings you dead mice.

They're going to pump this shit so full of advertising it's not going to be funny. And worse, it's going to be incredibly subtle and insidious. You're not even going to realize the responses are guiding you into certain purchases. Just gentle, reassuring advice that you should be buying shit.

This future sucks.

So now if you ask ChatGPT about what electric shaver to buy the answer will be sponsored? How nice.
I'd go darker than that. Think about how chatgpt is trying to position itself as a provider of medical advise.

Its easy to see how it will gently tune it's response from symptom to solution.

Cue rapid narrators voice "Chatgpt may cause death, seek real medical advice before buying what chatgpt recommends"

This is why normal people saying they are using ChatGPT to "research" which X to buy makes me so sad. Even if it is not a sponsored answer (yet) it is using data it found on random websites that it trusts are telling the truth. It is the dumbest game of telephone where the start of the chain is made up marking mumbo jumbo or straight fake, and each successive layer somehow speaks with more confidence and the end user thinks that because these LLMs have "access to all data from all time" that it means their answer will be correct.
Yes, there are enough people who still think this won't happen despite very conclusive recent history of this exactly happening.

Just like there are enough people who think if there were no car dealership/ salesman car would become cheaper as if car makers are full of morons who leave extra margins to buyers instead of themselves.

Or streaming service beats the shit of out cable in being cheap, ads-free and all in all better user experience. Once true, no longer.

Also funny that tips will go away when one visits self-serve in restaurants, coffee shops etc.

I guess a lot of folks are too drunk on "This time it will be different".

ChatGPT sends a lot of traffic to our app! In fact, most of our users come from ChatGPT.

My guess is that it’s because users talk to ChatGPT about our app, so it recommends it to more people?

Is your app AI-adjacent?
Boy, they just keep adding bigger and better reasons to never use ChatGPT. Good for them.
AGI has been achieved. Now let's grow ad revenue.
Nobody wants this.

Seriously what problem are we solving for here apart from Amazon trying to justify burning a giant pile of cash for OpenAI?

What problem is Rufus solving? Assistant that cannot do shit.
It’s allowing Andy Jassy to have something to tell investors to fend off criticism that Amazon’s internal AI efforts fell way behind the market on his watch. That’s the primary “problem” it solves for Amazon.

Other than that? Yes it’s total dog$hit.

> Nobody wants this.

Sure, but a billion people presumably do want to keep using ChatGPT for free.

That plan only works if they manage to drive Google et al out of business, and there’s zero evidence of that happening at this point. There’s no more ad $ coming to the table, just a battle for the same $.

If anything, things like Google putting really good AI right in search is making OpenAI’s efforts increasingly a hopelessly uphill battle.

Ads follow eyeballs. If people are spending time on ChatGPT then why won’t advertisers pay for ads over there?
Well, even payment is no guarantee that ads won't be injected into service. even these fucking movie theaters sho like 20-30 min of ads before showing movie people paid for.
Nobody wants the kind of advertising we get. If they did it would be optional.

There used to be optional ads. I used to pay money for adverts -- computer shopper and micromart in the 90s for example. There aren't now.

> Nobody wants this

"Oh. Well, then stop."

You would think Amazon would back off from being a "pilot" for a while.
too soon.

I'll be honest though, when I first read the headline I was trying to reconcile that with the crash. Took me a moment.

How epic, we supposedly have AGI if the marketing is to be believed, and what do these clever folks decide to do with their new omnipotent computer God that "solves coding" as one of the Anthropic drones so aptly said? More fucking advertisements. What a bright future we have ahead of us.
> Advertisers value Amazon for its troves of shopping and streaming data that can be used to improve marketing precision.

Thankfully I don't see their AI slop spam, not only due to ublock origin but because I have now made it my mission to support all who wish to eliminate ads on this planet. Since greedy corporations such as Amazon never learn, the counter-movement must help make ads totally obsolete. The sooner the industry accepts and learns that pestering folks is no longer acceptable, the better.

finally, a way for LLMs to hallucinate revenue
Ads are everywhere. I really think the 'fight against advertising' needs more of an aiki bent than it has. Advertising is such a massive juggernaut, we have as much hope of stopping it as a naked person has of stopping a freight train whose operator doesn't slow down for suicides.

But that image assumes we need to stop the momentum of this industry cold. personally, I believe that a course correction of a few degrees would be enough. Even that takes massive leverage though. I'd love to know who else is building in this space and what their approach is.