For nearly three decades, advertising has been the economic engine of the internet. It funds the search engines, social networks, publishers and countless digital services that billions of people use for free every day. The model has been relatively simple: publishers create content, platforms help people discover it, consumers provide their attention, and advertisers pay to reach them. It’s not a perfect system, but it’s been remarkably successful at creating and sustaining the internet as we know it for the last 30 years.
Since ChatGPT launched nearly four years ago, AI has been steadily reshaping how people discover and consume information. But that shift is now gathering momentum, and its consequences are beginning to ripple through the economics of the internet.
Over the past week, several new reports have caught my eye and highlighted just how significant these changes are now becoming:
OpenAI claims ChatGPT is now processing search queries equivalent to 20% of Google’s volume
Reddit has reported losing more than a million search referrals a week
Google is expanding advertising across its AI search experiences as growth in its traditional search advertising business slows
New research into ChatGPT advertising suggests that paid referrals generate 64% less revenue per visit than equivalent Google Search traffic.
I think these developments are all connected. Consumer behaviour is changing, established patterns of internet traffic and content distribution are being disrupted, and the advertising models that have sustained the internet for decades are struggling to keep pace.
But these changes aren’t happening at the same speed. Consumers are rapidly adopting new ways to discover and consume information, AI platforms are developing ways to monetise those experiences, and publishers are trying to adapt the economic models that fund the creation of information in the first place.
These three parts of the internet’s economy are becoming increasingly out of sync. And while the first two are moving quickly, the third is struggling to keep up. The result is a growing imbalance that could fundamentally change how the internet works.
An Economic Model Under Pressure
The World Wide Web was originally created as a way to share information. Over the last three decades, it’s evolved into somewhere we shop, work, communicate and entertain ourselves. But finding and consuming information has remained one of its most fundamental uses, and for much of its history that’s meant visiting the websites where the information was published.
AI has been fundamentally changing this. People are increasingly getting the information they need directly, without having to visit the websites where it originated.
For consumers, this is largely a good thing. It’s quicker, more convenient and often a better experience than navigating through multiple websites to find what you’re looking for. But for the businesses that create and publish that information, the implications are much more complicated.
Historically, appearing in search results meant an opportunity to attract visitors to your website. Those visitors could be monetised through advertising, subscriptions, product sales or other commercial activity. But when the answer is provided directly, the content creator can still be contributing to the experience without receiving any of the traffic or commercial benefit that would traditionally have come with it.
This is where the imbalance between consumer behaviour and the economics of content creation becomes particularly important. Consumer behaviour can change almost overnight as new AI capabilities become available, but businesses can’t adapt their economic models nearly as quickly. Publishers can lose traffic and revenue almost immediately, with fewer opportunities to generate advertising impressions or convert visitors into subscribers, while developing alternative ways to fund their content could take years. Meanwhile, the AI platforms capturing an increasing amount of consumer time and attention are still trying to work out how to monetise it effectively.
Google is in a somewhat different position. It already has an enormously successful advertising business, but it now faces the challenge of introducing AI experiences that change how people search without undermining the commercial model that makes search so profitable. OpenAI, on the other hand, is trying to build that commercial model almost from scratch.
But even if Google and OpenAI, or anyone else for that matter, eventually find ways to make AI advertising as effective and profitable as traditional online advertising, that doesn’t necessarily solve the problem for publishers. The advertising revenue generated by these platforms doesn’t automatically find its way back to the websites whose content they rely on.
This could create a significant imbalance. AI platforms could develop enormously successful advertising businesses while the publishers and content creators supplying much of their information continue to struggle. In other words, making AI commercially sustainable isn’t necessarily the same as making the internet commercially sustainable. Without new ways to compensate content creators, through licensing, revenue sharing or other arrangements, the gap between the two could continue to widen.
What Happens Next?
Publishers have been dealing with declining referral traffic for several years now, and many are already struggling to adapt their business models. The challenge is that AI is accelerating changes in how people discover and consume information, while the alternatives to the traditional advertising-funded model are still relatively limited.
Reddit’s recent consideration of restricting access to its content is an interesting example of how this might develop. By requiring users to log in, Reddit could strengthen its direct relationship with consumers, but potentially at the expense of making its content less accessible through the open web.
But there’s a potential problem with this approach. AI platforms themselves depend on access to reliable, current and diverse information. If content creators increasingly restrict access or can no longer afford to produce the same quantity and quality of information, the AI platforms relying on that content could eventually suffer too.
Licensing agreements and revenue-sharing arrangements might provide part of the answer, but they aren’t necessarily viable for everyone. Large publishers and established platforms have the scale and negotiating power to secure commercial agreements, but what about smaller publishers, independent creators and specialist websites? Collectively, they contribute an enormous amount of the information available on the internet, but individually they have very little leverage.
The longer-term consequence could be a fundamental shift in who can afford to create and publish information. Large publishers, established platforms and commercial organisations may be better positioned to adapt, while smaller publishers and independent creators find it increasingly difficult to sustain their businesses. Over time, that could leave us with an internet where information is produced by a narrower range of organisations, with fewer independent voices and specialist perspectives.
This makes finding alternative ways to fund content creation increasingly important. I’ve previously written about the emergence of Paid Organic, where publishers could generate revenue by making sponsored content available to AI platforms rather than consumers. It’s an interesting potential alternative to traditional advertising, although it introduces its own challenges around transparency and commercial influence.
And this brings us back to the different speeds of change. The commercial consequences for content creators are already being felt, while new business models that might compensate them are still being developed and tested. Even if the industry eventually finds a sustainable balance, there’s no guarantee that all the businesses supporting today’s internet will be able to adapt quickly enough.
The risk is that the open web gradually becomes less open, with more content sitting behind paywalls, registration requirements and commercial agreements, while an increasing proportion of the information we consume is delivered through a smaller number of AI platforms.
Finding a New Balance
There’s no doubt that AI is making the internet more useful for consumers. Being able to get the information we need quickly and easily, without having to navigate through multiple websites, is a significant improvement. And as AI continues to develop, the opportunities to make these experiences even better are enormous.
But we also need to consider the longer-term consequences. The internet depends on millions of businesses, publishers and individuals creating and sharing information. If the economics of doing so become increasingly difficult, we risk losing some of the quality, diversity and independence of the content that makes the internet so valuable in the first place.
The advertising-funded model that has supported the internet for decades was never perfect, and there’s no reason to assume it should continue unchanged. But as AI platforms develop new ways to monetise the information we consume, we need to ensure that the economic models supporting the creation of that information can evolve alongside them.
Otherwise, we could find ourselves with AI platforms that are enormously successful at delivering information, but an internet that's becoming less capable of producing it.
“The future is already here, it’s just not evenly distributed.“
William Gibson