AI Search Surfaces
The AI search surfaces are the products that answer a question in generated prose and attach links to some of their sources. Seven pages cover the ones that matter, and each is rated on how much its operator actually documents about source selection.
There are more answer products than there are answer engines. Several of the surfaces below share retrieval infrastructure: Microsoft Copilot is grounded in Bing, and the smaller AI assistants lean on a handful of the same indexes. Where two products retrieve from one index, being cited in one is a reasonable predictor of being cited in the other — and where they do not, it is no predictor at all.
What separates these pages is disclosure, not popularity. Google publishes a great deal about how AI Overviews are generated and how to leave them, and nothing about which eligible pages it chooses to cite. ChatGPT search documents its crawler behaviour and almost nothing about ranking. Perplexity makes citation the visible product and still does not explain selection. The verdict column below tracks that gap, not brand size.
Read each page for what its operator has committed to in writing, what independent measurement has found, and what the reporting surfaces — Search Console, Bing Webmaster Tools, referral headers — will and will not tell you about your own pages.
| Page | Evidence | What it is |
|---|---|---|
| AI Overviews | Documented | Google's generated answer block inside Search |
| AI Mode | Observed | Google's conversational search surface |
| ChatGPT search | Unsupported | OpenAI's web retrieval layer inside ChatGPT |
| Perplexity | Unsupported | The answer engine where citation is the product |
| Gemini | Unsupported | Google's answer app, outside Search and unreported |
| Microsoft Copilot | Observed | Bing-grounded answers with real publisher reporting |
| AI assistants and smaller engines | Unsupported | Claude, Meta AI, DuckDuckGo, Brave and Grok |