Video summary

Why an AI 'Death Spiral' Threatens the Internet

Main summary

Key takeaways

News and Commentary

Overview

The video argues that generative AI is reshaping how people find information online—particularly by reducing “click-through” traffic to publishers. This shift could evolve into an “AI death spiral” that threatens the health of the internet’s content ecosystem.

Key Points

1) Traffic is shifting away from links (“zero-click” search)

Westin explains that the internet has long depended on free or low-friction access through search, with publishers benefiting from ad-driven visits. But as AI becomes a new web “gateway,” search results increasingly answer questions directly—often without sending users to publisher sites.

The video uses Rand Fishkin’s clickstream research to support this:

  • Google’s share of searches leading to third-party websites has fallen:
    • from above 70% around 2011
    • to the mid-40s, and still declining
  • This reflects a zero-click trend.

Why it matters

  • Good for users: faster answers, less ad scrolling.
  • Bad for publishers: fewer referrals means less ad revenue, weaker subscription prospects, and slower audience growth.

2) AI deepens Google’s ability to disintermediate publishers

The discussion notes that Google has already answered questions directly in results for years (e.g., sports scores). AI “enhances” this capability—framed by Google as “AI Search, through and through.”

The implication: more information gets consumed on the search page itself, further eroding referral traffic.

3) News publishers face major revenue risk—especially smaller ones

Caitlin Petre highlights that publishers relying on Google referrals are seeing:

  • dramatic drops in traffic
  • corresponding declines in ad revenue
  • reduced potential for subscriber conversions

She also warns that AI may accelerate the elimination of news sources, echoing harms already seen in local news ecosystems under earlier business-model pressures.

4) Counterpoint: some large publishers are adapting—and profiting

Neil Vogel (People Inc.) offers a more optimistic perspective. Despite falling Google-driven traffic, he says his company is:

  • profitable
  • growing

His key claims include:

  • Audience reliance on Google Search dropped from ~75% to ~25%
  • People Inc offset the loss by diversifying distribution and building internet-based revenue

Their strategy includes:

  • Reducing dependency on search traffic
  • Expanding across platforms such as TikTok, Instagram, YouTube, and email
  • Licensing content to AI/LLM providers as an additional revenue stream

Vogel argues:

  • Licensing is “real money”
  • AI companies need power, models, and inputs, and publishers can monetize their role as content/training data providers
  • People Inc partnered with Cloudflare to block AI crawlers unless they pay, framing licensing as a way to “control our own destiny.”

5) The “death spiral” scenario: AI systems may run out of quality content

Vogel describes a likely (and tragic) feedback loop:

  1. AI reduces publisher traffic and revenue
  2. Some outlets—especially smaller/local ones—collapse or shut down
  3. Less high-quality content remains available to train AI tools
  4. AI companies respond by building internal publishing/creator systems—potentially resembling a “creator fund” model where AI firms pay creators to feed training/content pipelines

The result could be an AI-shaped internet:

  • content created primarily for and by AI tools
  • rather than an open, diverse web ecosystem

6) Broader societal concern

Vogel warns that declining journalism and content availability could mean:

  • fewer jobs
  • less robust information production He connects this to concerns about democracy.

The video also suggests AI companies should worry about content-supply problems: if reduced traffic undermines publishers, it can degrade training data quality and availability.

7) Pragmatic/optimistic view for business operations

Vogel closes with an internal business lens:

  • For People Inc, AI increases content output—described as producing significantly more content at roughly similar cost—by streamlining workflows.
  • He acknowledges the risk is real but argues companies can adapt by using AI operationally while developing new monetization and distribution strategies.

Presenters/Contributors

  • Westin (interviewer)
  • Rand Fishkin (author/entrepreneur; online search researcher using clickstream data)
  • Caitlin Petre (author and Rutgers professor)
  • Neil Vogel (CEO, People Inc.)

Original video