Google’s Head of Search Just Revealed How AI is Changing the Game: 5 Takeaways for SEOs

Is Search Dying? Absolutely Not. It’s Mutating.

Liz Reid the VP leading Google Search, recently sat down for a candid hour-long interview with Bloomberg’s Odd Lots podcast. This wasn’t a polished keynote; it was an unfiltered conversation about the trajectory of the world’s most used tool – Google Search.

Going through the full interview and independent research Here is what I learned, and what it means for anyone who depends on search traffic.

The old SEO playbook is breaking. The new one is built on intent, not keywords; on being a source, not just a ranker; and on understanding a fragmented, AI-powered search landscape.

Takeaway 1: The 20-Year Lie: AI in Search Is Nothing New

Let’s get one thing out of the way, that AI in search is something that arrived with ChatGPT. It has not.

“AI has been in search for many years in different forms,” Reid said. The timeline she laid out is worth paying attention to. Spell correction in the early 2000s? That was AI, even though, as she noted, “no one calls spellcheck AI anymore, but like it was at the time.” RankBrain arrived in 2015 and was later confirmed as the third most important ranking signal. Then came BERT in 2019, a transformer-based language model applied to 100 percent of English queries, which was arguably the largest change in search history until AI Overviews.

The point is not trivial. The team building Google Search has been managing the integration of AI into a live product used by billions of people for two decades. That institutional experience is something no competitor can replicate quickly. It is also why the “ChatGPT will kill Google” thesis that dominated late 2022 has not aged particularly well.

The current period, with AI Overviews generating summaries directly in results and AI Mode offering a conversational interface for complex queries, is more visible to users than earlier transitions. But it is not the first time the underlying search system has been rebuilt around new AI capabilities. It is simply the most visible.

Recent Key development everyone should aware of.

Year	AI Milestone	What It Changed
Early 2000s	Spell correction	First consumer-facing AI in search
2015	RankBrain	ML-driven query interpretation; later confirmed as Google’s 3rd most important ranking signal
2019	BERT (100% of English queries)	Transformer-based language understanding across all searches
2021	MUM (Multitask Unified Model)	Multi-modal, multi-language understanding
May 2024	AI Overviews launched	Generative AI summaries visible directly in results, 200+ countries
2025 onwards	AI Mode (conversational)	Full natural-language, multi-turn search interface

For a comprehensive map of this arc, DemandSphere’s tracker of 173 Google algorithm updates from 2000 to 2026 makes the two-decade journey visible in one place.

Takeaway 2: The Two-Decade Habit That Is Finally Breaking

For twenty years, we trained ourselves to compress our real questions into keyword shorthand. You wanted a restaurant for five people including a vegan and two kids, not too expensive, in a specific part of town on a Saturday evening. You typed “restaurants New York.” The gap between your actual need and your submitted query was enormous. It existed because we all learned, through experience, that search engines could not handle specificity.

That pattern has now reversed. Reid described it perfectly: “People stop talking just in keywordese as much, and they start expressing more of what they want. They tell you the real problem. They don’t take their need and translate it into what the computer understands. They try to give the computer their actual need and expect AI models to do the translation”.

This is not just about longer queries, though they are measurably longer. AI Mode queries run two to three times longer than traditional search queries. The real shift is that users have stopped doing the cognitive work of compressing their intent. The translation burden has moved from the user to the machine.

And the data backs this up. Google Ads data analyzed by Search Engine Land found that short queries of one to two words dropped from 42 percent of impressions in January 2025 to just 31 percent by June, while three-to-four-word queries grew significantly. That shift has almost certainly accelerated since.

What this means for you: Keyword research built entirely on short-form head terms is increasingly measuring behavior that fewer and fewer users are actually performing. Content structured to answer a whole problem, not just target a single keyword, is what performs in this environment.

Key pointers to consider.

• Optimise for full problems, not just keywords
• Build FAQ and conversational content that anticipates how people actually speak
• Structure content to answer layered intent: who, what, when, where, how much
• Use tools like Google Search Console’s query report to spot natural-language queries already reaching your pages

Takeaway 3: When (and Why) Google Decides to Show an AI Overview

One of the most practical questions is deciding when Google shows an AI-generated summary versus a traditional 10 blue links. The answer is not about the presence of a question mark. It is about intent.

Consider two searches. Someone types the name of a popular electronics brand. They likely want to navigate to that brand’s website, browse products, or find a store. An AI summary explaining what the brand does is unhelpful and unwanted. Now consider someone searching, “Why does my refrigerator make a loud humming noise only at night and how can I fix it?” That is a diagnostic need. The user wants an explanation, a set of possible causes, and a path to resolution. An AI Overview that synthesizes repair advice from multiple sources is genuinely useful.

Reid was also clear that AI Overviews are not shown simply because deploying AI is itself a goal. “We shouldn’t give you AI for the sake of giving AI. The point is for it when we think it adds value to people.” As models improve, the system can handle more cases without degrading quality, but the expansion is deliberately tied to capability, not ambition

And here is the part that should interest anyone whose traffic depends on informational content. As of early 2026, AI Overviews appear in roughly 47 percent of Google searches, with a significantly higher trigger rate of about 58 percent for informational queries. That is up substantially from the early days of the rollout and still climbing.

The practical implication is direct. Informational content that triggers AI Overviews is not doomed. It is the entry point to a citation. And being that citation matters enormously: websites cited within AI Overviews enjoy 35 percent higher organic click-through rates and 91 percent higher paid click-through rates compared to those not cited, according to a Seer Interactive analysis of more than three thousand queries.

Takeaway 4: One Search, Many Surfaces

A practical point that often gets missed is that Google is no longer a single interface. Users move between the traditional search results page, the AI Mode conversational interface, and the Gemini app, sometimes within a single session. These surfaces serve different tasks.

The traditional results page is built for navigation and transaction. Someone searching a specific bank name wants to log in, not read an AI explanation of what a bank is. AI Mode is for multi-step research, the kind of query where a user expects to refine, ask follow-ups, and iterate. Gemini handles creative and productivity tasks, rewriting, generating, analyzing a document, summarizing a long thread.

The important takeaway is that a brand’s content needs to be intelligible and authoritative across all these surfaces. A product page that is well-structured for a traditional search crawl may also need to be formatted in a way that an AI system can extract its key specifications and return them in a conversational answer. The same content serves multiple endpoints, and the optimization strategy needs to account for that.

Takeaway 5: The Future for Marketers and Publishers

The search box is not dying. It is becoming more capable of understanding what people actually mean. The marketers who succeed in this environment will be those who stop optimizing for keyword strings and start building content that answers complete human problems. They will be primary sources, not aggregators. They will optimize for being cited, not just ranked.

Keypoints:

Multiple surfaces require multiple strategies. A user in AI Mode asking a conversational research question is a different audience than a user typing a navigational query into the traditional search bar. Your optimization strategy needs to account for both, plus Gemini, plus whatever comes next.

Short-tail keyword strategies are decaying. Campaigns built around one-to-two-word keyword matching are encountering users who are now operating in a mode where they express full intent. The keyword and the intent are no longer the same unit. Content needs to be structured to answer whole problems, not target individual keywords.

Being cited beats ranking. A position-one ranking without an AI Overview citation is worth less today than it was two years ago. A citation inside an AI Overview drives meaningfully higher click-through rates. The optimization target is shifting from traditional rankings and traffic volume to authority and brand visibility within AI-generated answers.

Query volume is growing, not shrinking. The “expansionary moment” Reid described means more total questions are being asked than ever before. Some fraction of those carry commercial intent. New advertising surfaces are being built inside AI Overviews and AI Mode. The pie is bigger, even if the traditional click-through model is under pressure.

Final point to note that Google’s Search business just posted $60.4 billion in quarterly revenue, growing 19 percent year over year, with queries at all-time highs. Whatever disruption is coming, the evidence so far suggests that AI is expanding search rather than replacing it. AI is making search more human. If your content provides genuine, expert solutions to complex problems, you are positioned to win in this new era.

Watch the entire podcast here.