Google Downplays GEO as AI Search Quality Slips

Google has recently offered guidance on how content should be structured for AI-driven search, particularly pushing back against the idea that pages need to be broken into rigid “chunks”. While this advice may be technically sound, it arguably sidesteps the more pressing problems now facing search quality and visibility.

SEO And The Shift To AI-Driven Search

For many years, Google Search worked largely around keyword relevance, with PageRank extending this by factoring in links and anchor text. The launch of the Knowledge Graph in 2012 marked a turning point, as Google framed it as a move away from simple strings of text towards understanding real-world entities and relationships.

At the time, Google described this as the beginning of a new era of search – one that could better interpret intent and context in a more human-like way. That ambition is now being realised through large language models and AI-generated answers.

What Has Actually Changed For SEO

It’s fair to say that the foundations of Google Search are still in place. However, the way results are delivered has shifted dramatically. Instead of short answers or lists of blue links, users are now presented with long-form responses that attempt to address multiple related questions at once.

This has broken the long-standing SEO model of targeting a single keyword for a single result. With AI expanding queries behind the scenes, one search can now fan out into several related prompts, changing how pages are selected and surfaced.

Google’s Advice On Content Structure

In response to uncertainty among publishers, Google’s Danny Sullivan and John Mueller discussed content strategy on the Search Off The Record podcast. One key takeaway was their rejection of the popular idea that content should be deliberately split into small, self-contained “chunks” purely for AI consumption.

Many SEOs have assumed that because language models process information in segments, pages should be written in the same way. Google disagrees. Well-built web pages already contain structure through headings, lists, paragraphs, and semantic HTML. Artificially reworking content for machines risks undermining readability for humans.

Danny Sullivan was clear that Google does not want publishers creating separate versions of content for AI systems and traditional search. The emphasis, he said, remains on writing for people first.

Optimising For Machines Rarely Lasts

Sullivan also warned that tactics designed to satisfy specific systems tend to have a short shelf life. As search systems evolve, they increasingly reward content that is genuinely useful, well written, and human-focused. Techniques aimed solely at pleasing algorithms often fail to deliver long-term value.

This mirrors a familiar pattern in SEO history: shortcuts may work briefly, but strong fundamentals are what endure over time.

The Bigger Issue Google Isn’t Addressing

Where Google’s commentary falls short is in addressing the impact of AI search on traffic and referrals. Query expansion means fewer sites are being surfaced for a broader range of searches, reducing opportunities for specialist publishers to be discovered.

More concerning is the quality of the pages now being promoted in AI-generated results.

Expertise Is Being Pushed Aside

High-quality, expert content is increasingly hidden from default search views. In many cases, authoritative journalism and specialist publications only appear after users dig into secondary tabs such as “More” and then “News”.

As a result, AI Mode often highlights content that lacks credibility or subject-matter expertise.

When AI Results Miss The Mark

A simple, everyday search can now surface questionable sources. Examples include outdated personal blogs, generic social media articles, or retailer content that has little authority on the topic being discussed. These pages are often elevated above established publications that would normally be considered reliable.

The issue isn’t a single poor result, but a pattern that suggests quality signals are being overlooked.

Hiding The Best Content

Ironically, when users take extra steps to explore deeper search options, they often find the expert articles they were looking for all along. Publications with strong editorial standards still exist, but they are no longer front and centre.

GEO, SEO, Or Something Else?

Whether this shift is labelled GEO, AEO, or simply SEO evolving is largely irrelevant. What matters is that respected websites are losing visibility, and with it, sustainable traffic. The joy of discovering insightful, authoritative content through search is fading.

A Call For A Reset

Search has become cluttered with low-value results, while genuinely useful pages struggle for exposure. Many publishers are feeling the strain, and users are left sifting through content that offers little depth or originality.

Perhaps it’s time for Google to rethink how AI is integrated into search – restoring prominence to trusted sources and placing experimental features somewhere less dominant. Until then, concerns about structure and chunking feel secondary to the broader decline in search quality.