What if their AGI/ASI told them the road to Humanity betterment (because you know that was OpenAI initial mission, just in case we forgot) was through more efficient ads ? /s
AGI makes a lot better argument if everyone is in idiocracy.
IIRC, a part of the backstory of Idiocracy is the development of pervasive and good enough AI/automation, which enabled society to get that dumb without immediately collapsing.
If they need capital in the short term and were anywhere close to AGI, they could easily generate it, by spinning up their incredibly capable and intelligent agents to generate capital through doing useful work. They could also fire entire departments within their own org and replace them with agents, if they had something close to AGI.
Instead they are selling a dream to the gullible, and advertisements to go along with that dream.
This sounds awful. AI Chatbots will be everywhere Shopify is. I just see a lot of popups in our future with fake names and faces pretending to be people asking us how they can help us.
As usual with AI customer interaction, I have to ask who is liable for misleading advertising or unmet promises? The advertiser who isn't really in control of the chatbot, or OpenAI?
(Law on this goes all the way back to Carbolic Smoke Ball)
Liability... come on what's this pesky detail when it's an arm race with China (or whatever the weekly scapegoat is), don't you want the good guys to win? /s
Well to date when this has come up it’s been the company that put the AI in from of their customers that’s held liable in the first instance. See Air Canada case where an AI made up non existent policies that courts ruled Air Canada even had to honor.
Naturally then the company could turn around and sue the maker of the AI for damages and this is the bit that’s less tested to date.
But it’s clear the answer isn’t “nobody” is liable. So if this AI marketing widget makes up false claims about a product that’s going to be a big problem for companies that put an AI in front of their customers.
Tech companies have not had to answer to the law for a long time. We are now seeing that tech companies straight up break into each others servers and nobody is charged with any crime for it. I think it is safe to say that crime is legal for rich tech companies.
So who is liable? Nobody. This is a big tech company, to which laws do not apply.
Honestly I wish every ad had this. If I search for meal delivery services in my area and a sponsored result comes up, I would much rather ask an AI to trawl through the website to find out how much the meals cost, how much meat, protein, calories, etc are in them rather than do it myself
Yeah, but why would you trust the answers from a sponsored agent? Why would you assume it is giving you real, accurate, and complete information rather than selectively giving you what it thinks is most likely to make you click the "buy" button?
You can do that now. The problem with ads in an agentic environment is the AI would just ignore the best options and make up shit about worse options because a loss good restaurant with a big marketing budget paid it to say that. The whole idea of agentic workflows is only viable if the agent is truly independent and objective as best it can be.
I'm definitely more pro ad than the average HN user (I think they occasionally show me products I want to buy), but this seems bad and one step closer to "the LLM is biased by openai's financial interests."
In the example of a "sponsored agent" at the top, how many people are going to think "chat with us" is just a way to continue the conversation? How soon before openai removes the friction of clicking that entirely?
this is what openai said before: "Ads do not influence ChatGPT’s answers. Ads run on separate systems from our chat model, and advertisers have no ability to shape, rank, or alter ChatGPT’s responses. Ads are separate and clearly labeled." https://help.openai.com/en/articles/20001047-ads-in-chatgpt
Do they need to rewrite that section? Because in the video there's an LLM chatbot inside the chatgpt app writing answers on behalf of "Heirloom grocery"
Disagree. I pitched this exact model to Google Search when we first started making LLMs. The idea was the bots were separate clearly from the core model so you’ll know when you were having a branded experience or not. And it s useful beyond ads. It can be for customer service or MCP interaction with the service directly.
yeah, I can see it as useful in customer service, but they're adding these sponsored agents directly in the chatgpt app. the word "ad" or "sponsored" isn't visible after you click "chat with us" at 0:15 in the video, only "learn more about business chats" which is vague. and if you scroll down you won't even see that, if someone switches apps for 20 minutes are they going to remember that this chat window is an ad?
This is awfully unimaginative. I don’t see people using this. Where is the incentive to continue your conversation with the sponsored agent as opposed to, you know, the one you’re already talking to.
How else are they to monetize text outputs? This was always the plan and it's going to get much worse (or better depending on which side of the fence you're on). Google did the same thing when it controlled web indexation: ads, sponsored results and SEO. All of these things will be coming to agents: you will be monetized or you will be paying not to be.
They could stick to taking payment in exchange for service like some weird old-fashioned business, but I guess that doesn't get the sort of growth that justifies a trillion-dollar valuation.
I thought AGI was supposed to replace most knowledge workers and they'd pocket the savings that their customers enjoyed by not having to pay a work force. That's what their valuation is based upon.
I guess we’re about to find out if this helps put out the massive pile of burning cash that OpenAI is trying to deal with, but I know nobody that wants this.
OpenAI claims we’re near AGI and this is the best strategy they have? Oy vey!
This is a solution looking for a problem, built to satisfy the creator instead of serving the buyer. This is exactly what PG warns against doing. One would think Sam would know this too.
I doubt any consumer of ChatGPT wants ads, and even if they did, it would be a specific kind of ad. They specifically would not want ads if they cannot identify them as ads (like google has eventually done on the search page) or if they are not good ads.
Is this not just connecting you with a sales rep? If so, who wants this is literally every mainstream consumer.
AFAICT the flow is click ad then get an agent that is grounded in the official sales resources. This should mean the agent has the latest and canonical product catalog, offers, etc. so the users own agent doesn’t need to search on the web.
Fun ideas - can my main agent invoke these as a sub-agent, so I can get offers from multiple competitors? Can I loop them into a conversation to debate their products and observe which wins on my objective function?
Concerns - of course if the harness injects (or can be prompt hacked to inject) personal information then it’s a big problem.
OpenAI claims we’re near AGI and this is the best strategy they have? Oy vey!
Where did they say it’s the best? Obviously a short-term bridge strategy to drive “mundane” economic activity and juice the IPO. They are of course working on drug discovery and nuclear reactor plasma physics too. A good strategy has a portfolio of plays.
I mean, not to be rude to this and all sibling commenters, but obviously everyone? I feel like this comment has no basis in reality.
People want things cheaper and they tolerate ads. Those two things put together means everything gets ads and gets cheaper? YouTube exists for free and costs billions to run. Companies A/B test ad tiers and paid non-ad tiers, and the results are always the same.
They probably have the most amount of private amount of information about "you". Ads is a logical step for them, since subscriptions, at the current price, don't pay for the costs.
I think that’s the real plan. Grandma taps “find out more” and continues conversing with the helpful robot, unaware it has become the FurnitureCo sales robot.
Like Uber and Lyft showed in the last cycle, AI has now reached the milestone where the game shifts from "offer unreasonably cheap service by burning investor money and gain marketshare" to "shit we need to show a profit".
Or, now lets what the enshittification movement can do to this space.
Only a society which doesn't want anything, needs nothing, demands nothing can do without ads. While privacy invading ad tech is obviously a bad thing, ads - namely the thing used to introduce you and others like you to a new idea - is a good thing.
In the long run, the problem they'll run into with this is that the the people who have enough disposable income to pay for those plans are the people that advertisers will pay the most to show ads to. This is the same reason why the ads on daytime television are mainly for structured settlement cash-outs, drug class actions, and injury attorneys rather than for something that the viewer would have to pay money for.
The problem with ads in AI, is that they're incredibly creepy. People have told the apps all kinds of sensitive information that they never would have told a normal service. Obviously the advertisers are creaming themselves at that, but it's a new level of creeping.
It also allows them to subtly influence the messaging. It would be like if somebody tried to not bring out the Heinz at a family gathering. It just wouldn't be the same, because of the subtle messaging that has been inserted into the content. How could you trust them?
I think the level of invasiveness is probably closer to existing ads than you think. Not that that's a good thing, mind you, I think your disgust level here is probably pretty accurate to the level of personalized data that is currently being hawked by data brokers all over the internet.
Whats creepy to me is people telling all sorts of sensitive information to a chatbot and NOT expecting weird consequences. The level of cope is massive.
I agree with the general point on AI ads being a slippery slope and potentially creepy (though ChayGPT shows them as banners and if implemented honestly I don’t see anything misleading).
TFA is connecting you with an official sales rep after you click an ad. Maybe best to consider this a play to take over market share from the existing sales bots? (And thence likely support chatbots too.)
It’s okay (i.e. understand able) having ads on a low cost subscription. Would i not want to be (part of) the product myself, i’d willing to pay premium price - and i do
Tell me a story of when a service introduced ads and that ended up being a great thing for users ? Because I don't know any.
Instead I know many stories where ads poisoned and destroyed the service and turned the Internet into a shit show.
Oh, but in this case, dystopia is highly unlikely, right ?
i am not clear what exactly will these agents do? are they like ai sales rep for brand for talking to user if they want to explore the product further?
Looks like your average ad with a chat interface after you click it. For most online shopping, I consider this stuff annoying. Top of funnel is more sensible IMHO. Advise me on what product to buy, and help me make an informed decision, but I guess that is not where the ad dollars sit.
"Nuke the entire site from orbit ... it's the only way to be sure."
None of this would be necessary if they thought they were actually anywhere near AGI...hmm.
What if their AGI/ASI told them the road to Humanity betterment (because you know that was OpenAI initial mission, just in case we forgot) was through more efficient ads ? /s
Sounds like the plot of a Douglas Adam’s book
They'll have to find a smarter species to emulate first. My money's on the octopus!
Excellent way to put it.
It’s always he guys who steal the technology from the inventors that try to present the technology far beyond what it is.
Signed, that guy at google who did some stuff about this and had it taken away.
Turns out, just like the Turing test, its easier to dumb down intelligent expectations than to improve intelligence.
AGI makes a lot better argument if everyone is in idiocracy.
I wish more people could read and appreciate that couple of sentences.
IIRC, a part of the backstory of Idiocracy is the development of pervasive and good enough AI/automation, which enabled society to get that dumb without immediately collapsing.
Even if it's not necessary, it brings more money.
It’s pocket change compared to the money that could be made with real AGI.
This move tells us they don’t think they’re anywhere near AGI, directly contradicting their bombast and puffery on the subject.
Disagree, they need a lot of capital in the next couple years, and the more revenue they have today, the more data centers they can start building.
Also investors (particularly non-AGI-pilled) want to see revenue number go up, regardless of AGI. This juices the IPO.
If they need capital in the short term and were anywhere close to AGI, they could easily generate it, by spinning up their incredibly capable and intelligent agents to generate capital through doing useful work. They could also fire entire departments within their own org and replace them with agents, if they had something close to AGI.
Instead they are selling a dream to the gullible, and advertisements to go along with that dream.
This sounds awful. AI Chatbots will be everywhere Shopify is. I just see a lot of popups in our future with fake names and faces pretending to be people asking us how they can help us.
2036
Synthetic humans wearing the faces of your loved ones surreptitiously cross your path, and casually mention their favorite products and services.
This is one of the subplots of Good Luck, Have Fun, Don't Die. Great film.
That's pretty much most T1 support already. I feel like all I do is talk to bots.
I think we're already there with that stupid AI customer support widget every obnoxious website has now.
As usual with AI customer interaction, I have to ask who is liable for misleading advertising or unmet promises? The advertiser who isn't really in control of the chatbot, or OpenAI?
(Law on this goes all the way back to Carbolic Smoke Ball)
Liability... come on what's this pesky detail when it's an arm race with China (or whatever the weekly scapegoat is), don't you want the good guys to win? /s
It cannot be the AI vendor, according to what they say they are purely passive, think of them like a very large harness for their AI models.
It’s not the advertising companies of course, as you said they don’t control the system.
And it cannot be the models, because they don’t have personhood.
So, I guess it’s on nobody? Funny how those things happen to work, uh
/s
Check out https://en.wikipedia.org/wiki/The_Unaccountability_Machine on this topic, available at all good bookstores near you.
Well to date when this has come up it’s been the company that put the AI in from of their customers that’s held liable in the first instance. See Air Canada case where an AI made up non existent policies that courts ruled Air Canada even had to honor.
Naturally then the company could turn around and sue the maker of the AI for damages and this is the bit that’s less tested to date.
But it’s clear the answer isn’t “nobody” is liable. So if this AI marketing widget makes up false claims about a product that’s going to be a big problem for companies that put an AI in front of their customers.
Tech companies have not had to answer to the law for a long time. We are now seeing that tech companies straight up break into each others servers and nobody is charged with any crime for it. I think it is safe to say that crime is legal for rich tech companies.
So who is liable? Nobody. This is a big tech company, to which laws do not apply.
Honestly I wish every ad had this. If I search for meal delivery services in my area and a sponsored result comes up, I would much rather ask an AI to trawl through the website to find out how much the meals cost, how much meat, protein, calories, etc are in them rather than do it myself
You can already do that without ads.
The point of ads is that one of those services will pay openai so that their services will be more prominent than the others.
No you can’t. Not without MCP
Most agents can search the web nowadays
My guy executing tasks like navigating door dashes app and getting a full basket price reliably pretty much needs MCP.
Sure you can have some agents try to do this via web but realistically most fail.
Yeah, but why would you trust the answers from a sponsored agent? Why would you assume it is giving you real, accurate, and complete information rather than selectively giving you what it thinks is most likely to make you click the "buy" button?
You can do that now. The problem with ads in an agentic environment is the AI would just ignore the best options and make up shit about worse options because a loss good restaurant with a big marketing budget paid it to say that. The whole idea of agentic workflows is only viable if the agent is truly independent and objective as best it can be.
I'm definitely more pro ad than the average HN user (I think they occasionally show me products I want to buy), but this seems bad and one step closer to "the LLM is biased by openai's financial interests."
In the example of a "sponsored agent" at the top, how many people are going to think "chat with us" is just a way to continue the conversation? How soon before openai removes the friction of clicking that entirely?
this is what openai said before: "Ads do not influence ChatGPT’s answers. Ads run on separate systems from our chat model, and advertisers have no ability to shape, rank, or alter ChatGPT’s responses. Ads are separate and clearly labeled." https://help.openai.com/en/articles/20001047-ads-in-chatgpt
Do they need to rewrite that section? Because in the video there's an LLM chatbot inside the chatgpt app writing answers on behalf of "Heirloom grocery"
Disagree. I pitched this exact model to Google Search when we first started making LLMs. The idea was the bots were separate clearly from the core model so you’ll know when you were having a branded experience or not. And it s useful beyond ads. It can be for customer service or MCP interaction with the service directly.
yeah, I can see it as useful in customer service, but they're adding these sponsored agents directly in the chatgpt app. the word "ad" or "sponsored" isn't visible after you click "chat with us" at 0:15 in the video, only "learn more about business chats" which is vague. and if you scroll down you won't even see that, if someone switches apps for 20 minutes are they going to remember that this chat window is an ad?
It’s not an ad. It’s a business agent. The entrance is an ad.
This isn’t that hard to understand.
This is awfully unimaginative. I don’t see people using this. Where is the incentive to continue your conversation with the sponsored agent as opposed to, you know, the one you’re already talking to.
The fact that it requires a user action to initiate is better than the previous solution, but that's a pretty low bar to have
How else are they to monetize text outputs? This was always the plan and it's going to get much worse (or better depending on which side of the fence you're on). Google did the same thing when it controlled web indexation: ads, sponsored results and SEO. All of these things will be coming to agents: you will be monetized or you will be paying not to be.
They could stick to taking payment in exchange for service like some weird old-fashioned business, but I guess that doesn't get the sort of growth that justifies a trillion-dollar valuation.
It has for Anthropic. Consumer vs enterprise though I guess.
Without the pesky training costs or stock compensation taken into account, yes.
By making more money from inference subscriptions than they lose on compute, training, and other costs?
I thought AGI was supposed to replace most knowledge workers and they'd pocket the savings that their customers enjoyed by not having to pay a work force. That's what their valuation is based upon.
Here I foolishly thought that that was the job of ChatGPT itself.
This is the lie advertisers tell themselves to be able to sleep at night.
Ads are cancer on content. No one wants them. They only serve to make the presentation of useful content worse.
It's funny that at first OpenAI didn't want to add ads, but if they are adding them, they will be as dystopian as possible.
Who actually wants this?
I guess we’re about to find out if this helps put out the massive pile of burning cash that OpenAI is trying to deal with, but I know nobody that wants this.
OpenAI claims we’re near AGI and this is the best strategy they have? Oy vey!
Nobody wants this. We only need them to offer propaganda as a service and the enshitification will be complete.
Isn't that primarily what they offer anyway? So "subscription required" would check that box.
Who wants any ads?
Investors
Sam Altman. It is his last resort, he said some time ago (which is not true, regulation/"slowing down the pace of AI" seems to be the latest).
Could be exploring old and new ways of monetization.
This is a solution looking for a problem, built to satisfy the creator instead of serving the buyer. This is exactly what PG warns against doing. One would think Sam would know this too.
OpenAI's shareholders, and only them.
I doubt any consumer of ChatGPT wants ads, and even if they did, it would be a specific kind of ad. They specifically would not want ads if they cannot identify them as ads (like google has eventually done on the search page) or if they are not good ads.
Advertisers, for one.
People that don't want to pay for everyday AI, which is most people.
Is this not just connecting you with a sales rep? If so, who wants this is literally every mainstream consumer.
AFAICT the flow is click ad then get an agent that is grounded in the official sales resources. This should mean the agent has the latest and canonical product catalog, offers, etc. so the users own agent doesn’t need to search on the web.
Fun ideas - can my main agent invoke these as a sub-agent, so I can get offers from multiple competitors? Can I loop them into a conversation to debate their products and observe which wins on my objective function?
Concerns - of course if the harness injects (or can be prompt hacked to inject) personal information then it’s a big problem.
Where did they say it’s the best? Obviously a short-term bridge strategy to drive “mundane” economic activity and juice the IPO. They are of course working on drug discovery and nuclear reactor plasma physics too. A good strategy has a portfolio of plays.
Any basis for believing this?
If it helps the lower tiers of ChatGPT continue being free, I'd support it.
I mean, not to be rude to this and all sibling commenters, but obviously everyone? I feel like this comment has no basis in reality.
People want things cheaper and they tolerate ads. Those two things put together means everything gets ads and gets cheaper? YouTube exists for free and costs billions to run. Companies A/B test ad tiers and paid non-ad tiers, and the results are always the same.
They probably have the most amount of private amount of information about "you". Ads is a logical step for them, since subscriptions, at the current price, don't pay for the costs.
the only person who ever interacts with these will be the person who accidentally clicked on them. Surely they know this?
I think that’s the real plan. Grandma taps “find out more” and continues conversing with the helpful robot, unaware it has become the FurnitureCo sales robot.
Soon, everybody will live in Truman show.
I dont think its that bad. Another opportunity for businesses
Like Uber and Lyft showed in the last cycle, AI has now reached the milestone where the game shifts from "offer unreasonably cheap service by burning investor money and gain marketshare" to "shit we need to show a profit".
Or, now lets what the enshittification movement can do to this space.
So you can pay $8 a month and see ads while using the already pared down offering.
I think pushing back with cancellations is the only way forward.
I wish ads never existed.
The future is open source, no subscriptions, no purchases, 100% free and community supported.
The sooner software gets to this point, the world will be a much more better place rather than everything being a subscription.
ChatGPT is already free to use? What, do you expect them to keep providing this without any return?
Besides, who will pay for the compute of LLMs in your utopian future?
Only a society which doesn't want anything, needs nothing, demands nothing can do without ads. While privacy invading ad tech is obviously a bad thing, ads - namely the thing used to introduce you and others like you to a new idea - is a good thing.
So you want software from people for free, hope you share also your work power with others for free.
I the end someone pays for open source and often it's some developer with his spare time.
Yes to open source, but yes to money for the developers.
Found this:
https://help.openai.com/en/articles/20001047-ads-in-chatgpt
For now.
Just like cable tv, and Netflix, and Amazon Prime Video, and....
In the long run, the problem they'll run into with this is that the the people who have enough disposable income to pay for those plans are the people that advertisers will pay the most to show ads to. This is the same reason why the ads on daytime television are mainly for structured settlement cash-outs, drug class actions, and injury attorneys rather than for something that the viewer would have to pay money for.
The problem with ads in AI, is that they're incredibly creepy. People have told the apps all kinds of sensitive information that they never would have told a normal service. Obviously the advertisers are creaming themselves at that, but it's a new level of creeping.
It also allows them to subtly influence the messaging. It would be like if somebody tried to not bring out the Heinz at a family gathering. It just wouldn't be the same, because of the subtle messaging that has been inserted into the content. How could you trust them?
I think the level of invasiveness is probably closer to existing ads than you think. Not that that's a good thing, mind you, I think your disgust level here is probably pretty accurate to the level of personalized data that is currently being hawked by data brokers all over the internet.
Whats creepy to me is people telling all sorts of sensitive information to a chatbot and NOT expecting weird consequences. The level of cope is massive.
Yeah, I'm sure people do it, but I'm having a hard time picturing why...
I agree with the general point on AI ads being a slippery slope and potentially creepy (though ChayGPT shows them as banners and if implemented honestly I don’t see anything misleading).
TFA is connecting you with an official sales rep after you click an ad. Maybe best to consider this a play to take over market share from the existing sales bots? (And thence likely support chatbots too.)
Nice so we're getting closer to Black Mirror in real life.
It’s okay (i.e. understand able) having ads on a low cost subscription. Would i not want to be (part of) the product myself, i’d willing to pay premium price - and i do
What’s wrong with this choice ?
I guess you can pay OpenAI for the privilege of hosting a RAG database rather than having their agents trawl your site.
So you cant control the agents you have, and now they will be sponsored so at least they make money off the chaos?
Here we go again..
Tell me a story of when a service introduced ads and that ended up being a great thing for users ? Because I don't know any. Instead I know many stories where ads poisoned and destroyed the service and turned the Internet into a shit show.
Oh, but in this case, dystopia is highly unlikely, right ?
i am not clear what exactly will these agents do? are they like ai sales rep for brand for talking to user if they want to explore the product further?
truth seeking interface + influence seeking commercial interface = conflict of interest
Looks like your average ad with a chat interface after you click it. For most online shopping, I consider this stuff annoying. Top of funnel is more sensible IMHO. Advise me on what product to buy, and help me make an informed decision, but I guess that is not where the ad dollars sit.
I am surprised nobody brought up Accelerando yet.