AI in the Wild | Your Chat Bot Has a Sponsor
You are the product
We all know the phrase that goes along the lines of “if you’re not paying for the product, you are the product” It’s from the 1970s and was a criticism of TV but more recently was used to critique Facebook. Truthfully, it doesn’t feel so bad when it’s Facebook or Google and I’m the product.
But when it’s my AI chat bot? Those conversations are very different to Facebook likes or random search queries on Google. They’re often a lot more personal, a lot more detailed and at times real-feeling conversations.
So, when in an interview from October 2024 last year OpenAi CEO Sam Altman said “I think about ads as a last resort for our business model”, it sounded like we wouldn’t be the product. Then February this year saw ChatGPT start serving ads to logged in US users on the free and $8 Go tiers of ChatGPT.
The ad-supported internet won
We had this debate around the internet over twenty years ago and the ad supported internet won out. All the world’s major websites rely on adverts as their primary source of income. In short the internet of today is a ‘free’ product funded by surveillance capitalism.
Do we want our AI to be the same? Surveying our chat bot conversations in order to deliver relevant ads?
Anthropic is clearly saying no ads (Disclosure: I’m a paying Anthropic subscriber). They even took out an advert for the Super Bowl half time show taking shots at OpenAI and their stance on adverts. In one, a man asks his AI how he can get closer to his mother and gets delivered an ad for a dating site called “Golden Encounters”.
But somebody has to pay. In the early 2000s ads won because people would not pay for the internet. The AI industry has a window to establish a different norm before the economics force the same outcome.
Trust is the currency
Perplexity, an AI aggregator, tested ads during 2024 and 2025 but abandoned the idea back in February. The reason according to their CEO was simple: user trust. Users need to believe they are getting the best possible answer, and ads undermine that confidence.
Now if we’re worried our search result may not be the best, we can live with it. But if our AI chat bot is a poor performer, that’s potentially a bigger issue.
OpenAI has said that they will “not optimize for time spent in ChatGPT. We prioritize user trust and user experience over revenue.”.
We’ve seen this movie before
Sure, but we’ve seen this movie before. Some will remember Google’s motto “Don’t be evil” coined around 2000 and included in their IPO prospectus in 2004.
They quietly dropped it in 2018 and went full evil as we found out during discovery for the Alphabet anti-trust case in 2023. Emails surfaced that showed Google had deprecated search results to boost revenue.
In an email from 2019, Ben Gomes, the former head of Search at Google, commented that “I am getting concerned that (revenue) growth is all we are thinking about”.
The problem was that search revenue was lagging and they needed more income from ads. But the search engine was too efficient. We never left the first page or needed to do a follow up search. So Google tweaked the algo and presto, the results got worse and suddenly we’re onto page 3 of a search query and then redoing it looking for the answer. We’re spending longer searching across mor pages. Ad revenue goes up and we get a markedly worse user experience.
Older readers will remember in the early days of Facebook they promised we’d control our data and vote on policy changes. Well that’s long gone.
My Instagram feed is pretty much accounts I don’t follow interspersed with adverts. The simple chronological time line option is buried and not the default. The solution to make it better for me is easy for Meta, but bad for revenue. So Metas answer to me is suck it up.
The cash burn problem
But how big is the actual funding problem for AI companies?
Currently AI companies are all at the cash burning stage. They’re raising billions from investors, but they’re spending many more billions. OpenAI is reported to be on track to lose $14billion this year. This can’t last forever. Remember back in the early days of Uber the trips were dirt cheap? Basically rich Silicon Valley investors were sponsoring me to get around Johannesburg. But that had to change and Uber is no longer cheap and AI chat bots simply can’t be free forever.
Now the different business models matter here. Anthropic is very much a B2B operation and selling their API to large Fortune 500 companies earns a decent fee. In turn they can then offer me a discounted AI chat bot.
OpenAI also does some B2B but has a lot more consumer focus. They have 900million weekly users, but only around 5% pay . Perhaps because the $20 monthly fee does not pay for much chatting with your AI chat bot.
So what else is there?
Cross-subsidising is one option that Microsoft and Alphabet will likely use. You pay for an annual 365 live subscription, now you get co-pilot included. Your new Windows machine comes with co-pilot, paid in part by the laptop manufacturer. Google searches already come with AI content right up at the top and they can further bundle it within the paid Google Workspaces offering.
Commissions is another. If my AI agent buys something on my behalf (something AI companies are pushing hard as a large part of the future for them), they could earn a commission. But here the risk is that they’ll then buy the highest commission paying product rather than the cheapest or best product Again we’re back at trust.
Apple may have an answer: hardware margins. Apple makes a ton off your fancy iPhone or MacBook and they then use some of that profit margin to offer you bundled AI as they’re proposing to do using Alphabet’s Gemini. We pay them margin, they (reportedly) pay Alphabet $10billion a year for Gemini and we get it for ‘free’. This will work well for top end products, but not the budget phones.
All have risks, some worse than others.
It comes down to trust
AI is not anything like search or doom scrolling social media. It is deeper, more meaningful, truthfully more useful and as such is really is about trust. Surveillance capitalism doesn’t work here.
We need to decide which paying options we like and reject the outright ad model. I’m happy to pay for Anthropic and get Gemini for ‘free’ on my Apple products. But I will actively not use AI that includes ads as their revenue model.
Founder Just One Lap
AI in the Wild
AI in the Wild is a regular column from Simon Brown.
AI is everywhere and only getting better. Record capital raises and valuations, and competing LLMs are all fun, but meaningless to our every day lives. This column will focus on how is it impacting us in the real world.
5th | 15th | 25th every month
