The woman who's run Google Search for two decades just admitted the quiet part out loud: AI is eating her product, and she's the one holding the fork.
The Summary
- Liz Reid, Google's head of search for 20+ years, sat down with Bloomberg to discuss how LLMs like ChatGPT and Claude are fundamentally changing how people query the internet
- Google is integrating Gemini directly into search via "AI overviews" — a move that creates internal tension between AI utility and the ad revenue that still pays the bills
- The rise of LLMs means users are bypassing traditional search bars, forcing Google to cannibalize its own dominant product or watch competitors do it first
- Reid's challenge: navigate "internet slop" (AI-generated garbage clogging the web) while maintaining traffic that drives Google's core business model
The Signal
Google built a $200 billion-a-year business on the premise that people need to click through search results to find answers. Now the company is embedding Gemini directly into search, delivering answers without the click. This isn't innovation. This is controlled demolition.
Liz Reid has been at Google for over 20 years. She's seen the company survive mobile, social, and voice. But LLMs are different. When users can ask ChatGPT or Claude a question and get a coherent answer without ever seeing a blue link, the entire search advertising model starts to wobble. Google's response: AI overviews that surface Gemini-generated summaries at the top of search results.
"More and more people are getting their online info through AI, effectively bypassing the search bars of old."
The tension is obvious. Every AI overview that satisfies a user without a click is revenue Google doesn't collect. But if Google doesn't offer it, users will go elsewhere. Reid is managing the transition from "ten blue links" to "one AI answer," and she's doing it while the company still relies on those links to pay for everything.
Here's what makes this harder: internet slop. As LLMs flood the web with synthetic content, the quality of searchable information is degrading. Google has to:
- Filter AI-generated garbage from real sources
- Train Gemini on a web increasingly polluted by other LLMs
- Maintain user trust when the internet itself is less trustworthy
The "search in an AI world" framing misses the point. Search isn't adapting to AI. Search is being replaced by AI, and Google is trying to make sure it's their AI doing the replacing.
The Implication
If you're building in the agent economy, watch what Google does here, not what it says. AI overviews are a test case for how incumbent platforms will integrate agents without blowing up their business models. The ones who figure out how to monetize answers instead of clicks will own the next decade. The ones who don't will become infrastructure.
For knowledge workers, this is your canary. When Google — the company that literally defined "search" — starts delivering answers instead of links, your job is next. The question isn't whether AI will replace information retrieval roles. It's whether you'll build the agents that do it or be replaced by someone who did.