Google's Liz Reid on Who Will Own Search in a World of AI

What It Covers

Google's VP of Search Liz Reid explains how AI overviews and AI mode are reshaping search behavior, why the click-through advertising model remains viable despite AI-generated answers, and how Google distinguishes between Gemini and Search as separate but complementary products serving distinct user intents.

Key Questions Answered

• •AI Overview Deployment Logic: Google does not show AI overviews on every query — it deploys them only when user signal data confirms they add measurable value. Queries where users simply want to reach a specific destination, like typing "Wikipedia" or a brand name, receive no AI overview because the intent is navigation, not information synthesis.
• •Query Expansion as the Core Business Case: AI overviews drive revenue not primarily through click monetization but by expanding total query volume. Users mentally filter out questions they deem too time-consuming to search. AI lowers that friction threshold, generating entirely new query categories — particularly in non-English languages where web content is sparse but LLMs can bridge the gap.
• •Ad Revenue Resilience Mechanism: Search ads appear on fewer than 25% of queries, meaning most AI overview queries were never monetized to begin with. For commercial queries, an AI-generated answer does not eliminate purchase intent — users still need to select a merchant, making ad placement downstream of AI answers a viable and potentially better-targeted format.
• •Gemini vs. Search Product Differentiation: Users self-segment across Google's products by task type. Informational and research queries trend toward Search and AI Mode; creative and productivity tasks like text reformatting trend toward Gemini. AI Mode specifically attracts users expecting multi-turn, conversational follow-up rather than single-answer lookups, mirroring how Maps and Search coexist without collapsing.
• •AI Recruiting Interview Evolution: Google's technical interview process is actively adapting to AI coding tools, shifting assessment focus from rote problem-solving toward evaluating a candidate's fluency with AI-assisted workflows. Interviewers now consider whether candidates can critically direct AI tools rather than simply produce code, with the benchmark itself changing every three to six months as tool capabilities advance.