We have Tim Berners-Lee Thanks to the World Wide Web. But these days, the British computer scientist’s creation is at risk, thanks to the rise of generative AI. As large language models (LLMs) increasingly generate information from the Internet, fewer people are visiting websites for that source material — an evolution that could “disintegrate” the Web’s advertising-based business model. According to Berners-L
“If LLMs are reading it and people aren’t reading it, then we have a problem with the business,” said Berners-Lee, who invented the World Wide Web in 1989, speaking at the FT Future of AI Summit yesterday (November 5). “We have to replace it with something else.”
A shift toward AI-generated summaries is already underway at companies Google Integrate AI across their search engines. Share of Internet users likely to click on traditional search results Cut in half when offered a Google AI overview summaryAccording to a recent Pew Research Center report, which also found that only 1 percent of surveyed users visit the links cited within that summary.
Reshaping the Internet’s business model isn’t necessarily a bad thing, says Berners-Lee, who notes that some users have become “disgusted” with the ad-based model. Highly personalized ads in particular can “freak out” people and make them feel like they’re being watched, he added.
Despite AI posing a threat to the web’s current economic base, Berners-Lee is not opposed to the technology. In fact, he is the co-founder of an AI startup himself: Inrupt, which is developing a chatbot called Charlie. The bot draws on personal data to provide customized responses, allowing users to control which platforms can access their information.
Still, Berners-Lee warned that the rise of AI could have serious consequences for the quality of information that relies on humans. The technology is “terrible in the sense that so much of our life on the web is based on reading pages in one way or another, and that part is taken out of the mix,” he said. “If humans only use AI, why wouldn’t humans read blogs?”
He worries that users may accept LLM output at face value and avoid fact-checking through sites like Wikipedia. The online encyclopedia has already seen Human visitor numbers are down 8 percent this year. It is responsible for getting answers from AI-powered search engines instead of users.
Fewer human visitors to websites also means less new information being written—creating long-term risks for LLMs, which require new data for training. One possible outcome, Berners-Lee suggests, is that AI systems could eventually create their own components. “Perhaps you end up with a society where LLMs play the role of writer as well as reader,” he said.
Such a future isn’t as dystopian as it sounds, Web’s creator added. “There’s an assumption that AI-generated stuff is kind of hogwash,” says Berners-Lee. “But there will also be good AI stuff.”
