Google says AI should help SEO, not replace user value

Google is drawing a clear line for agencies: use AI to do better work, not to crank out more thin pages. In a May 1 Search Off the Record episode with Martin Splitt and Nikola Todorovic, who is director of software engineering at Google Search, the message was that AI should help businesses analyze data, research competitors and sharpen the value they deliver to users.

Todorovic’s point was practical, not dreamy. There is no magic fix, no single roadmap and no shortcut that guarantees success in AI-driven search. The baseline still looks stubbornly familiar: sites, products and platforms have to earn attention by being useful. For agencies, that is the useful signal buried under all the AI hype. It means AI belongs in research, content planning, competitive analysis and QA, while editorial standards still have to hold the line on originality, accuracy and actual user benefit.

The bigger search backdrop makes that advice harder to ignore. Google said in 2025 that AI Overviews were available in more than 200 countries and territories and more than 40 languages, and that the feature drove more than a 10% increase in usage for the kinds of queries where it appears in major markets such as the United States and India. Google has also described AI Mode as using query fan-out, a technique that breaks a question into subtopics and runs many searches at once. That is a strong hint that longer, more conversational queries are becoming a bigger part of how people search and how pages get surfaced.

Google followed that with a May 6 update saying it was improving AI Mode and AI Overviews to make it easier to find relevant websites, deep insights, original content and direct links. That matters for publishers and agencies because it pushes against the fear that AI search is only about keeping users inside Google. The company is saying it still wants to send people back to the open web, especially when a page offers something specific enough to be surfaced and clicked.

The editorial takeaway is straightforward. Build AI into the workflow, but tie every use of it to a quality check, a usefulness check and a client expectation check. Google’s own guidance points to a market where the winners are not the teams publishing faster, but the teams using AI to produce sharper research, better pages and more defensible value.