If you search for a lasagna recipe, you will likely visit a food blog,You read the ingredients, learned a family cooking technique, and your visit generated ad revenue for the creator. Today, you ask a search engine for a recipe, and an AI synthesizes a summary. You get the instructions, the search platform gets monetizable traffic, and the creator gets nothing. The internet’s fundamental value exchange is breaking down.
The Clickless Internet
The traditional web relied on a simple contract: traffic in exchange for content. Now, platforms are intercepting the audience. According to market research firm SparkToro, over half of all Google searches in the U.S. and Europe in 2024 ended without a single click. An April 2025 analysis by SEO platform Ahrefs showed that when an AI overview appears, clicks to top-ranked organic results drop by an average of more than a third. “Answer engines” are keeping users on their interfaces and starving publishers of revenue.
The Model Collapse
As creators lose financial incentives, they publish less free content. This leads to a compounding problem identified in a 2024 Nature study: “model collapse.” Because human-generated data is drying up, AI systems increasingly train on synthetic data generated by other AI models. The process mirrors making a photocopy of a photocopy. Over time, the nuanced human details fade, and the models amplify errors and biases, leaving behind a degraded, sterile information base.
The Data OPEC
With the open web shrinking, publishers are locking high-quality human data behind paywalls. Hamilton Mann, writing in Noema Magazine, compares this shift to a “Data OPEC.” Just as an oil cartel restricts supply to control markets, a few powerful platforms and rights-holders could soon monopolize access to premium training data. This gatekeeping restricts open innovation, raises costs, and concentrates informational power among a select few.
Artificial Integrity
Sustaining the web requires a shift to what Mann calls “Artificial Integrity.” This framework embeds verifiable, machine-readable citations directly into AI outputs, ensuring structural credit for creators. It also demands a new economic model where the act of an AI citing a source triggers compensation, even if users never click through to the original site. Finally, tech platforms must be required to reinvest a portion of their revenues into maintaining the open web. Without these financial and architectural guardrails, AI systems will eventually run out of the reliable information they need to function.
ref: https://www.noemamag.com/the-ai-powered-web-is-eating-itself/
