Keep pulling the thread on Liz Reid.
The shift towards users seeing answers directly in AI-generated summaries could negatively impact Google's advertising business, which relies on clicks to external links.
The introduction of AI Overviews has led users to ask longer, more conversational questions in Google Search, including query types they might have previously taken to a standalone LLM.
A key metric Google uses to evaluate changes to its search product is whether the changes lead people to initiate search sessions more frequently.
Google views AI as an "expansionary moment" for search because it can unlock a large volume of user questions that were previously unasked due to perceived difficulty or time commitment.
Google Search shows advertisements on less than a quarter of all search queries.
Google has observed that the use of AI Overviews correlates with users submitting meaningfully longer search queries.
Google has observed an increase in natural language queries from users interacting with its AI-powered search features.
With features like AI Overviews and AI Mode, Google observes that users are shifting from "keyword-ese" to describing their actual needs in natural language.
The rise of AI is perceived as a potential threat to Google's core search business.
Google transformed its search ranking systems several years ago using AI models like BERT and MUM, which were based on early Transformer architecture breakthroughs.
Google's user data indicates that people want a search experience that integrates both AI-generated answers and traditional web links, rather than choosing one over the other.
Google has observed that AI Overviews lead to fewer "bounce clicks," where a user clicks a link and immediately returns to the search results page.