Search engine optimization and AI: what's real and what's hype

More people now ask an AI assistant the questions they used to type into Google, and the assistant answers with a short list of businesses, products or websites. A new industry has grown up around getting your business onto that list. You may see it sold as AIO, answer engine optimization, generative engine optimization (GEO) or "AI SEO." The pitch is the same under every name: ordinary SEO no longer works, and your website needs a new kind of optimization before ChatGPT, Google Gemini, Claude, Copilot or Perplexity will recommend it.

People really are searching differently. Whether that means your website needs a new kind of optimization is a separate question, and Google answered it in May 2026. Its guide to generative AI search says its AI features run on the same ranking and quality systems as ordinary search, names AEO and GEO as terms in circulation, and concludes that optimizing for generative AI search "is still SEO."

AI search is not another list of Google results

A conventional search engine takes your query, scores the pages in its index and returns a ranked list of links. An AI assistant can do more with the same question. It can interpret what you are asking, run several searches, pull information from several sources and assemble one answer.

Google calls this "query fan-out": one question becomes a set of related searches, each covering part of the subject. Google's AI features also use retrieval-augmented generation, or RAG, which means the model looks up current pages in the search index while it writes the answer instead of relying only on what it learned in training. OpenAI describes the same pattern for ChatGPT.

Ask an assistant for "a good, affordable carpet cleaner near me that can handle wool rugs" and it may search separately for local carpet cleaners, for wool rug cleaning, for reviews and for service areas, then combine what it finds. That is a real change in how information is retrieved. It is not evidence that your pages need to be written in a special language for machines.

There is no single AI index

A common assumption is that every AI assistant repeats whatever ranks first in Google. They do not. Google's AI features use Google's index. ChatGPT has its own search crawler, OAI-SearchBot, and OpenAI says it also works with other search providers. Anthropic runs separate crawlers for model training (ClaudeBot), for search (Claude-SearchBot) and for pages a user asks Claude to open (Claude-User). Each company ranks results its own way, and none publishes the formula. OpenAI says its results are ranked on multiple factors meant to surface relevant and reliable information, and that placement is not guaranteed.

So when someone offers you a universal formula for "ranking in AI," ask which AI. There is no single index and no single ranking system to rank in.

Training a model and appearing in AI search are different things

An AI model learns general knowledge from a large body of text during training, and that knowledge is fixed until the next model is trained. Live search is separate: the assistant retrieves current pages at the moment you ask. For a business website, live retrieval is the part that works like SEO. You cannot engineer a page so that a fact about your business is permanently stored inside every future model. You can make sure your pages are open to search crawlers and say something worth retrieving.

OpenAI's setup shows the difference. It runs GPTBot to collect content that may be used for training and OAI-SearchBot for ChatGPT search, and you can block one while allowing the other. That choice is the subject of the third article in this series.

What about special AI optimization techniques?

This is where the marketing gets ahead of the evidence. Google's 2026 guide goes through the techniques being sold for generative search one by one. For Google's AI features you do not need AI-readable files, special Schema.org markup, content chopped into short "chunks," pages rewritten for AI or mentions manufactured around the web. An llms.txt file has no effect on Google Search visibility either way.

Other AI services may support other technologies, now or later. The point is to be skeptical when a tactic is presented as necessary because the letters "AI" are attached to it.

What actually helps

The list is familiar. Your website should be open to legitimate search crawlers. Its pages should say plainly what you do, where you operate and what makes you relevant. Your content should answer the questions customers actually ask. Your business information should match across your website, your business profiles and the directories that list you. Reviews and other third-party references tell an AI system how other people regard you. Structured data helps search engines read certain kinds of information, though Google says none of it is required for its AI features.

Above all, your content has to add something. Google's guide puts its weight on what it calls "non-commodity content": information that comes from your own expertise and is not already repeated across thousands of other websites. That matters more now that generic content costs nothing to produce. A plumbing company publishing "10 signs you need a plumber" adds almost nothing to the web. The same company explaining the drain problem it keeps finding in houses built in its area in the 1970s publishes something that exists nowhere else. The second article is useful because of what the business knows, not because of how it was formatted.

AI visibility is measurable, but the tools are new

Bing Webmaster Tools now reports when your pages are cited in AI answers across Bing and Copilot: which pages, which queries retrieved them and how citations trend. Google Search Console has a report for impressions in Google's generative AI features. These reports show you something that did not exist two years ago. They do not reveal a hidden optimization formula.

Be skeptical of guarantees

Nobody has ever been able to guarantee the top organic position in Google, and AI recommendations leave even less room for the promise. OpenAI states that placement in ChatGPT search is not guaranteed. Google says no third-party tool has access to its internal ranking or AI systems. You can improve the conditions that make your website discoverable and useful. You cannot dictate what an independent system recommends.

So, is "AI SEO" real?

AI search is real, and so are the new crawlers, citations and reports that come with it. What has not been shown is the need for a separate optimization discipline with its own content formats and ranking rules. The defensible strategy is the plain one: a technically sound website, a clear statement of what you do, original and useful content, accurate information everywhere else on the web, and open access for legitimate search systems. Good SEO is becoming more important as search changes, not less.

The next article looks at what happens when someone asks an AI assistant to recommend a business, and why the answer is assembled differently from a page of Google results.

In this series

1. SEO and AI: what's real and what's hype (this article)
2. How does an AI assistant decide which businesses to recommend?
3. Should you block AI bots from your website?
4. How to make your website visible in AI search without chasing AI SEO tricks