Sir Tim Berners-Lee doesn’t think AI will destroy the web

November 10, 2025·55 min·5 clips

Tim Berners-Lee explains why TikTok's algorithms are designed to keep you scrolling forever.

Tim Berners-Lee, creator of HTML and HTTP protocols that power the World Wide Web, discusses the current state and future of the internet with host Nilay Patel. Berners-Lee is co-founder and CTO of Inrupt and author of new memoir "This Is For Everyone," a phrase from the 2012 London Olympics opening ceremony. The conversation centers on whether new information ventures would still choose open web platforms over closed ecosystems like TikTok or YouTube. Berners-Lee observes that markets combined with networks inevitably produce monopolies, leading to Chrome dominating browsers, Google dominating search, and single players controlling social networks and marketplaces. He distinguishes between the web's declining role as an information platform and its strengthening position as an application platform, though AI threatens to disrupt both by becoming the primary interface layer. The original "This Is For Everyone" vision emphasized digital sovereignty and peer equality, which Berners-Lee believes platforms have eroded through addictive algorithms and data control. Platform CEOs argue they've democratized creative tools, but Berners-Lee counters that TikTok's addictive design and closed ecosystem contradict genuine user empowerment. At the web's inception, Berners-Lee convinced competing companies to adopt unified standards by arguing one web would grow exponentially while fragmented webs would fail individually. He doubts this collaboration could happen today with AI companies, though suggests figures like Demis Hassabis at Google DeepMind and potential government or philanthropic initiatives might create a CERN-like institution for AI development. Berners-Lee worries that AI agents browsing on behalf of users could destroy the advertising revenue model that funds much of the web infrastructure. His company Inrupt develops "Charlie," an AI assistant that accesses user data stored in personal data wallets using the Solid protocol, giving users control while enabling more effective AI assistance than systems relying solely on external data. The conversation reveals a fundamental tension: users claim to value privacy but consistently choose convenience, immediately using services like Instagram despite privacy concerns. Berners-Lee acknowledges market forces alone won't achieve data interoperability, suggesting regulation may be necessary, though he notes the Atlantic divide between Europe's regulatory approach and America's current deregulation push under Trump. He sees his semantic web vision—machine-readable data enabling new applications—paradoxically achieved through AI's brute-force extraction of training data, though acknowledges the extractive unfairness perceived by content creators. The discussion's tone is candid and reflective, with Berners-Lee neither fully pessimistic nor naively optimistic about the web's trajectory. His technical explanations remain accessible while addressing complex architecture questions about local versus cloud-based AI inference and data storage. The conversation suits audiences interested in internet history, technology policy, AI's impact on information ecosystems, and debates about user sovereignty versus platform control in digital spaces.

As heard by us

Berners-Lee makes the AI-web debate concrete by asking who an agent actually works for.

Nilay Patel's interview with Sir Tim Berners-Lee treats AI as a practical test for the web, less apocalypse than architecture. Berners-Lee argues that an assistant only becomes useful if it can work with a person's own data, held in a wallet, and if it has a duty to act in that…

Read the full review in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Decoder with Nilay Patel