For twenty-five years, the web has rested on a tacit agreement. You publish content for free, search engines index it, and in return they send you visitors. Those visitors pay for your work, through advertising, subscriptions or reputation. This pact was never signed anywhere, but it has funded a huge share of what you read online. AI is now breaking it, without really having decided to.
Ask an AI a question today, and you get an answer. Not a list of links to explore: an answer, directly, synthesised. It is infinitely more convenient. It is also, potentially, the end of the business model that has funded the web as we know it. Let's explore this tension, without catastrophism or naivety.
The mechanism, in one sentence
When an AI answers you using a site's content, you no longer visit that site. The content creator still provides the value, but no longer receives the counterpart. This is called clickless search: the question gets its answer without any visit being generated.
The phenomenon is established enough that Google has rolled out, in its publisher tools, a view dedicated to AI-generated impressions, i.e. the occasions where content was used to produce an answer. Telling detail: this view currently shows neither click data nor associated queries. You can see your content was used, without knowing what you get out of it.
Why this isn't just a media problem
The temptation would be to see this as a concern for newspapers and bloggers. That would miss the point. The web is not limited to media: help forums, developer tutorials, association documentation, recipes, manuals, comparisons, educational resources. This vast library was built by people, many of whom were paid indirectly through the audience they attracted.
If that audience disappears, the question becomes simple and uncomfortable: who will write tomorrow's content? This is not a moral question about fair compensation, it is a mechanical question about production. An ecosystem whose revenue source is removed ends up producing less.
Here is the twist. AIs need quality human content to be trained and to stay up to date. If they dry up the economy that produces that content, they are sawing off the branch they are sitting on. And the web then risks filling up with AI-generated content, which other AIs will in turn read, in that impoverishment loop we described in our article on model collapse. It is not just the web that has an interest in the pact holding: it is also AI.
The other side's arguments, which do exist
For honesty's sake, we must present the opposing thesis, because it is not absurd.
First, this prediction has been made several times before. The death of the web was announced with the arrival of mobile apps, then with the dominance of social networks. The web transformed rather than disappeared. It is plausible that it will transform again.
Second, a large share of the traffic captured by search engines concerned trivial questions, like unit conversions or opening hours, on which nobody was building a solid business model. Losing that traffic is not dramatic for quality content.
Third, a licensing market is emerging. Several AI companies now pay publishers to use their content, and it is precisely the absence of such deals that fuels the ongoing lawsuits against Google and others. Payment at the source could replace payment per visit. The problem is that this market structurally favours the big publishers able to negotiate, and leaves out the blogger, the association, the forum, i.e. a large part of what makes the web rich.
What will probably happen
The most likely scenario is neither the death of the web nor the status quo, but a recomposition in two movements.
On one side, generic content, easily synthesised, will lose most of its traffic value. Pages that answered simple questions no longer have a reason to be visited, and will gradually disappear.
On the other, what cannot be summarised will gain in value: original investigation, lived experience, verifiable expertise, community, a direct relationship with a reader. An AI can summarise a product review, it cannot have held that product in its hands. This is a shift analogous to the one we described for the developer profession: the middle compresses, the ends spread apart.
It is likely that the direct relationship will become central again. Newsletters, communities, subscriptions, everything that bypasses the intermediary to connect a creator to their audience without depending on an algorithm. The early web worked much like this, before search engines and social networks inserted themselves.
So, will AI kill the web? Probably not. But it is killing a specific business model, that of content funded by traffic. What comes after will depend on our collective choices: the rules we set on paying the source, and our individual habits. Every time you click to read the full article rather than settling for the summary, you vote, very modestly, for someone to keep writing it.