How AI is breaking the internet (and what to do about it) | Matthew Prince

Episode Summary

Executive Summary: Matthew Prince argues that AI answer engines are undermining the web’s traffic-based business model, threatening journalism and open-web content. He says this disruption could also enable a better system that compensates original knowledge, local reporting, and new ideas, potentially via AI licensing and a new recognition model for creators.

Main Topics: The internet’s traditional traffic economy (Priority: 5/5): Prince explains that for decades the web has been financed by traffic routed through search, clicks, ads, and subscriptions, with Google serving as the central mechanism for indexing and monetizing content. AI as a platform shift (Priority: 5/5): He frames AI not as a tool but as a fundamental interface change from search engines to answer engines, where users get direct responses and stop clicking through to publishers. Collapse of publisher traffic (Priority: 5/5): Prince cites data showing it is becoming far harder to drive traffic to content, which he argues threatens the economics of journalism, research, and the open web. A new model for rewarding original knowledge (Priority: 4/5): He proposes that AI companies should pay for content that fills gaps in human knowledge rather than rewarding inflammatory, low-cost content designed to maximize clicks. Examples of creative monetization (Priority: 4/5): Prince uses Spotify and his hometown newspaper to show how technology can create revenue for creators and local journalism through licensing and demand for unique content. Power struggles among AI and search firms (Priority: 4/5): He argues Google’s entrenched advantage and aggressive stance toward training data are slowing the emergence of fair compensation for creators. The 'Big Cheese Award' concept (Priority: 3/5): Prince imagines a prize for the entity that most advances human knowledge, symbolizing a shift from click-driven metrics to knowledge creation.

Key Arguments: The internet’s long-standing economic model is traffic: content is created, users click through, and publishers earn via ads, subscriptions, or commerce. Google historically made this system work by indexing content and monetizing search traffic, but AI answer engines bypass that pathway. Direct-answer AI experiences reduce clicks and therefore cut off the revenue that funds journalism and research. Publisher traffic has already fallen sharply, with reported declines ranging from 30% to 90%, indicating systemic disruption. AI companies still need journalism and original reporting because they require new knowledge to improve their systems. The future model should reward content that fills gaps in collective knowledge, not content that inflames attention to generate clicks. Technology can increase creator earnings, as seen in music streaming and in local-news licensing to AI firms. Google’s dominance in web crawling and search gives it an advantage over newer AI firms and complicates fair compensation for creators.

Data Points: Traffic harder to obtain vs. Google of old: 5.2 times harder - Prince’s measurement of how much more difficult it is to generate traffic from the same amount of content compared with earlier Google search behavior. Traffic harder to obtain vs. OpenAI: 750 times harder - Comparison cited by Prince for answer-engine traffic displacement. Traffic harder to obtain vs. Anthropic: 32,000 times harder - Prince’s extreme example of reduced referral traffic from AI systems. Publisher traffic decline: 30% to 90% decline - Reported range of traffic losses media companies are experiencing. Time since Napster disruption: 23 years ago - Prince uses the music industry’s piracy crisis as a precedent. iTunes-era pricing: 99 cents per song - Steve Jobs’s proposed model for paid digital music consumption. Spotify creator payouts: Over $10 billion last year - Used to show that tech platforms can expand creator revenue. Spotify query monetization example: 40 million euros a year - A Danish songwriter earning from filling unfulfilled Spotify requests. Local voter turnout: From the low 20% to over 60% - Prince’s Park Record ownership example showing local news impact on civic participation. AI-crawling comparison: OpenAI sees 1 page; Google sees 3; Microsoft sees 4; Anthropic sees 5; xAI sees 21 or 22 - Illustrates Google’s crawl/indexing advantage over competitors.

Pivotal Quotes: "Traffic has been the currency of the internet" — Matthew Prince: Core explanation of the web’s historical business model. "We have something that's closer to what I would call an answer engine" — Matthew Prince: Describing how AI changes user behavior by providing direct answers instead of links. "We could call it the Big Cheese Award" — Matthew Prince: His proposed symbolic prize for the biggest contribution to human knowledge.

Implications: If AI keeps diverting users away from links, journalism and open-web content may lose their funding base. But licensing, better AI compensation, and rewards for original knowledge could create a healthier internet economy.