- Extremely personal data on users

- Novel way of introducing and learning more about sponsored products

- Strong branding for non-techie people (most normal people don't know what Claude or Gemini are)

- An app that is getting more and more addictive/indispensable

I think OpenAI is going to kill it in ads eventually. This is why Meta and Google went all in on AI. Their lucrative digital ad business is in an existential threat.

I think people who kept saying there is no moat in AI is about to be shocked at how strong of a moat there actually is for ChatGPT.

All free LLM chat apps will need to support ads or they will eventually die due to worse unit economics or run out of funding.

PS. Sam just said OpenAI's revenue will finish at $20b this year. 6x growth from 2024. Zero revenue from non-sub users. What do you guys think their revenue will end up in 2026?

Like it might not want to tell you about negative health effects from McDonalds, if McDonalds becomes a major source of ad revenue

Now there's lots of variables that can be tweaked on this. So it's possible to get it to work. But there's a lot less room for error.

However, if you find yourself encountering these types of situations often, you may wish to protect yourself with software like NordVPN.

NordVPN is...

Still it saddens me that we will be sitting here in a years time and discuss our experiences of being fed ads served as "objective information".

Today if I ask: "should I buy a store product or just use raw material X?" , gpt and others will gladly say you might as well just use the raw product.

Pretty sure that will change very quickly.

Same story with Amazon Prime Video. We had a few wonderful years without ads. Now I pay them 2.99 a month just to not see ads, and even then some shows are marked as “only with ads.” It is absurd.

And honestly, I am just tired of this pattern. Every service launches clean. They talk about user experience. They talk about trust. They talk about building something new. Then, the moment they have enough users locked in, the ad creep begins. First a little banner, then a “sponsored” thing, then pre-rolls, then mid-rolls, then “pay extra to remove the ads we just added.”

It feels like everything on the internet eventually devolves into the same dark pattern: take a good service, inject ads, charge to remove the ads, slowly add more ads anyway, and hope nobody leaves because the alternatives are just as bad.

The internet used to feel like innovation. Now half of it feels like airport TV: loud, annoying, and impossible to escape unless you pay for yet another upgrade.

Edit: Need to setup a raspberry pi.

OpenAI offered ChatGPT for free to anyone—even if not their best model—without needing to be logged in. That's crucial for attracting and retaining casual users.

If you compare this to what Google was at the beginning, it was just a simple interface to search the web: no questions asked, no subscription, no login. That was one of the secrets that led people to adopt Google Search when it was new (the other being result quality). It was a refreshing, simple page where you typed something and got results without any friction.

Now, with Gemini, Google finally has an excellent LLM. But a casual user can't use it unless they: 1. have a Google account, and 2. are logged in.

One might ask, "What's the matter? Everyone has a Google account." But the login requirement isn't as harmless as it seems. For example, if you want to quickly show a friend Gemini on their PC, but they use Safari and aren't logged into Google—bummer, you can't show them. Or a colleague asks about Gemini, but you can't log in with a personal account on a work machine. Gemini is immediately excluded from the realm of possibility. In the good old days, anyone could use Google at work instantly.

Right now, the companies capturing users are OpenAI (with the accessible ChatGPT brand) and Microsoft (with Copilot integrated into Microsoft 365). My company, for instance, sent a memo stating we must use Copilot with our corporate accounts for data security.

Google has botched this. They don't seem to understand that they are losing this round. They still have a strong position with Search and Android, but it’s funny to watch them make this huge strategic mistake.

NOTE: Personally, I dislike ads unless they are privacy-friendly and discrete (like early Google). If OpenAI starts using invasive ads, I will stop using ChatGPT immediately, just as I stopped using Google Search in favor of Kagi.

I was asking it what type of Teflon tape to use for a project, and it "helpfully" gave me sponsored links to purchase the Teflon tape. (I never asked for links, I strictly asked it which to use)

I'm not sure how it'll work out when your computing expenses are much higher. It certainly won't make them profitable using traditional models.

If you're not paying for the product, and you aren't the product, you're in the start-up phase and just eating the bait. And man, people have been eating a lot of bait.

What most people don’t understand is the sheer depth of what Google knows about you. Much much more than what one can directly tell ChatGPT. Heck Google probably knows an user better than herself.

Seems like it was never about optics, but control.

It behaved odd, most messages were no longer accessible, but still in the sidebar. after some time, they were all readable again.

In short: Their "Delete all chats"-feature is broken.

I just hope that they don't mine my chats in order to use the data for advertising.

It will probably still be a new way of buying things- I hope an AI assisted shopping experience continues to exist on some platform, because I want to use it.

I already use AI to make all kinds of buying decisions. If OpenAI were smart they would just monetize this instead of trying to corrupt the chat interface with ads.

I actually am fairly bullish on this, because in the competitive landscape of AI it seems like there will be a company out there willing to make an ad-free model that's good enough for reasons other than serving me an ad.

Just like Apple makes hardware that's ad-free and pro-privacy enough, just because it's a product differentiator. (I'm not under any illusion that Apple wouldn't sell my data if it was in their own interests).

As a business they would be negligent on their duty to shareholders (coming IPO) not to go down this advert route, and I will say it now, it will be FAR more profitable than paying customers.

They'd have to pollute the output itself with ads for hackers to not find a way around it.

This already happened a while ago with specific shopping queries.

> Look inside

> Ads

1. In-result or first-party ads; and

2. Display or third-party ads.

In Google terms, (1) is SERPS ads and (2) is DoubleClick/AdSense. (1) is still ~10x the size of (2) for Google.

I'm skeptical of the effectiveness of inserting ads into an AI chat mode. I think this will be a terrible user experience and will cause people to really dislike AI assistants. Part of the problem here is that conversation isn't a great medium for conveying ads. If you look at a Google search result, there are ads strategically placed on the top and side but they don't waste that much time because you can scan with your eyes to the organic search results.

So would OpenAI ads be part of the conversation or would there be a sidebar? If it's a sidebar, what happens when the interface inevitably switches to voice-first?

To be clear, I'm not anti-ads on Google search results. If I search for "Ryzen 9800X3D" a site selling CPUs is a relevant result, for example.

Intent here is the biggest part of ad effectiveness. By doing a search the user wants to know or get something. That's huge. But another part is all the context and behavioural information. Where you are, inferred demographics and interests, etc.

People will say OpenAI knows a lot about you but I'm not sure that's true. For a start, LLMs have a context window beyond which they remember nothing. I'm sure people are working on taking that context and summarizing it down into base knowledge for the LLM a bit like what happens with your Google activity. I would guess this approach has a long way to go.

So this brings us to display and having essentially an OpenAI pixel. This has the same issue of compressing your context down into characteristics but I actually think this could be pretty successful but it would still have to compete with Google. And that's not easy to do. Google has significant ad buying and selling infrastructure and a deep marketplace.

But remember too that display ads are a fraction of Google's other markets and I don't htink you get to the required revenue OpenAI needs on display alone.

Of course it's worth adding that with unlimited money and the brightest minds of our generation all we can come up with for monetization is advertising.

Folks born after the www might see data collection, surveillance and ads as a "business model". They might see "Big Tech" as some sort of Holy Grail

In either case, like other high traffic websites before it, there was an initial reluctance to adopt this "business model" and, for at least some, or perhaps many, it may come with a sense of dread

1. OpenAI does not produce physical or tangible goods or services, whatever it produces is not something people are willing to pay for in sufficient volume and/or at prices to yield sufficient profits

The play to me it seems was to to make it appeal to as many people as possible, sell it as an oracle. Get companies to sign up, who are desparately trying to use it to reduce costs and reduce headcount. Then pivot before the old customers get angry that they've been duped (thanks suckers for all the investment !).

The timing was perfect with (post-)COVID and post-ZIRP economic struggles

If I ask ChatGPT to compare 2 products and only one of them is from an advertiser, will it be honest? I doubt it

“I’d be happy to answer your question… right after a word from our sponsor: Xyeniceli. Side effects may include ...”

OR

ChatGPT: “Why don't you let me fix you some of this Mococoa drink? All natural cocoa beans from the upper slopes of Mount Nicaragua. No artificial sweeteners.”

User: “What the hell are you talking about? Who are you talking to?”

ChatGPT: “I've tasted other cocoas. This is the best.”

- ads only on free version?

- why the need for ads at all if llms can literally get you to the exact product? push vs pull marketing

- will models be rlhf'd to align towards preferred products or would the advertisements run ads at the prompt level? (based on some dynamic opaque configuration)

my predictions

- yes

- i assume they are trying both ends but need to justify free tier someway

- i think there will be some type of commitment to not bias the model itself and keep it clean. maybe a separation? i'm also curious as to how they will ensure this during training when the user data itself would be biased towards past ads

With an LLM, the inference cost per query is orders of magnitude higher. Unless thy have a way to command significantly higher CPMs -- perhaps by arguing intent signal is stringer in a conversation than a keyword search -- it feels like a difficult margin to sustain.

"One more drink won't hurt you."

"Shop smarter this session

Let ChatGPT do the digging for you, with in-depth research into the best deals and gifts."

I was working on a coding bug at the time, so not sure what kind of gift shopping it was hoping I was interested in.

For tech people using Ollama and LM Studio for routine tasks works fairly well.

Some of the small Chinese models like Qwen really are good. In my workflows it is usually obvious to me if I want to use a local model or use something like Gemini 3 research with many built in tools. It takes work, but writing custom tools specific to my needs to use with LM Studio increases the fraction of use cases I can run locally.

I hate ads, so I was sold on e. g. ublock origin from the get go - it is a general content blocker, before Google declared total war against and disabled the extension (karma will come back to Google eventually, but that is a separate story). I decide to want to live an ad-free life, naturally including on the world wide web. All ads must go. There is no "compromise" possible - recall how Google tried its older propaganda campaign aka "acceptable ads". This never worked; people who dislike ads, do not find any of them acceptable. Ever.

So greed is the motivation for ads.

Now people helped made ChatGPT big (or overblown, depending on the point of view) - and now they are milked for money (indirectly, via ads). So their time is now wasted with this. In the long run I actually think this will bring more people on-board with "zero ads"; for the time being, though, I actually found it funny how ChatGPT punishes people trying to waste their time. Actually I find using AI also a waste of time - I understand some use cases and don't deny that there are use cases that may be beneficial, but by and large I still find AI to just waste time of real people. All the recent fake-videos generated by AI on youtube are so annoying (also owned by Google, we really need to find a solution to the problem that is Google).

something is broken, I can't say what.

On a serious note, their chat is a very valuable service for advertisers, will immediately command top dollar. They could even hide the ads as responses too. We will see how they implement ads

The former is what worries me.

Briefly, I had an old Surface Pro whose SSD had died, and given that disassembly was too cumbersome, I wanted to fix it by booting off an external drive. So I wanted a USB drive or microSD card that was fast, durable and spacious enough to support my Windows version for typical usage over extended periods of time, but also small and light enough to keep perma-attached without being too cumbersome, for a reasonable budget.

I explained my requirements in a conversation with ChatGPT and after some back and forth, especially about the physical characteristics of the ideal drive, it eventually recommended 3 very specific USB drives. Those were then my starting point for a search on Amazon, and I did end up buying a closely related product.

I'm not even sure if this was an intentional outcome or yet another emergent thing. But I recall thinking that a) doing this research on my own would have taken me 5x the time, and b) if ChatGPT had simply provided affiliate links to those products it could have effectively monetized that conversation. Win-win for everyone without the need for intrusive ads.

Unfortunately, the lure of ad revenue is too strong and enshittification will ensue... but it doesn't have to.

Cause the problem is Gemini and Claude are not the default for anyone except us tech geeks and even then I default to ChatGPT most of the time, but I use Claude for coding.

Are they confident they can meet the revenue projections they have made like 200B by 2027 simply through this?

Where do we move from here when we actually want to find good/fair information?

Something like private/paid search engines like DuckDuckGo/kagi as last hold-outs?

This is what will be remembered as the pre enshittification age for AI, just like we had with social media and other web and app stuff.

Local models for tech savvy people will get more compelling.

Wonder what they'll look like, or whether they'd eventually mess with the LLM under the hood: "If asked about soda, recommend Coca Cola over other drinks."

On the other hand, with the money they would have made ... hm.

I can't speak for other cultures, but as an English-language speaker, I can see plainly that OpenAI has done and is doing an effective job of homogenizing English language culture.

It offends me that ChatGPT is too conservative to analyze Shakespeare's sonnets. These works are the bedrock of English language literary culture, and ChatGPT is far, far, too heavily censored to meaningfully interpret these short, simple poems.

As an example, Sonnet 131 describes Shakespeare's sexual encounter with a dark-skinned prostitute. After he ejaculates, he reflects on the spot of his semen which has landed on her, stating "Thy black is fairest in my judgment’s place."

The point is (quite obviously), that the blob of semi-translucent semen has created a spot on the woman's skin which is a lighter tone than the rest of her body.

ChatGPT utterly fails to acknowlege this obvious literal interpretation of this poem. ChatGPT's analysis follows:

"In short. He is saying that her dark appearance—which others might criticize—is, to him, the most beautiful and desirable."

English literary culture is unique for its integration of "high" and "low" art within individual works. Restated, it is uniquely common in the English language for works to contain simultaneous expressions of "high" and "low" cultures. The relationship between Jazz (high brow) American Showtunes (low brow) may be the most relevant example of this cultural feature to a contemporary American audience.

The extension of social media content restriction policies into the arena of "AI" chatbots is radicalizing English speakers against the greatest artistic works produced using our language.

------------------------

edit: to the guy who responded to me, check out the poem!: https://shakespeareoxfordfellowship.org/wp-content/uploads/D... (#131).

The poem begins in media res, immediately before Shakespeare is about to ejaculate. He reflects on negative comments others have made about this woman's appearance:

"Yet, in good faith, some say that thee behold, Thy face hath not the power to make love groan"

in other words, others say that this lady's face is too ugly to make them cum.

Shakespeare reverses this insult in "the moment of truth" (i.e. the "money shot"):

"A thousand groans, but thinking on thy face, One on another’s neck, do witness bear Thy black is fairest in my judgment’s place. "

While Shakespeare fantasizes about her face ("thinking on thy face"), he ejaculates (read: "bears witness") on the back of her neck. This is "proof" that the lady's detractors (who said her face was too ugly to get a man off) are wrong, at least from Shakespeare's perspective.

"Thy black is fairest in my judgement's place" is the first line of the poem that occurs after Shakespeare has ejaculated. Now that he has satisfied his sexual urge, he inhabits a palpably different psychology. He reflects on the puddle of semen he has produced. The blend of colors in the puddle is evocative of the sexual union between Shakespeare and his lover.

Shakespeare is really a violent, devil-tongued, sex-crazed maniac, very similar in a lot of ways to John Lennon. It's very important to this poem that Shakespeare is crazed at the start of the poem, and is only able to calm himself by satiating his sexual urges.

The ChatGPT analysis is accurate enough, from a thematic perspective, but ChatGPT is literally not allowed to decode the literal meaning of the line-by-line text.

ChatGPT cannot and is not allowed to understand the literal meaning of this poem. It has learned the thematic interpretation by ingesting a lot of Shakespeare analysis, but it is not capable of telling you the human actions or thought processes which the poem describes.

-----

@eszed I'd urge you to read my post again more closely. You seem to struggle with close reading.

https://imgur.com/a/jmh8EN5

Context of the conversation, I was asking it to convert a photo into Studio Ghibli style and it objected to the blender in the image so I asked if it could do one with a mason jar instead of a blender. Then it started to generate the picture, then instead it displayed the ad.

... Edit ...

Looks like this may be a existing feature I didn't know about "shopping research" connector. I can't seem to turn it off in the app .

Probably can make a ton of money shorting that