Google Search is moving generated answers into a more visible role. With AI Overviews, some searches now begin with a summary produced by generative AI, followed by supporting links that may or may not be immediately visible.
The change matters because it alters a basic habit of the web: searching, scanning links, choosing a source, and reading. AI Overviews are designed to save time, but they also raise questions about verification, source visibility, and how much users should trust a fast answer.
What Google AI Overviews do
AI Overviews appear at the top of Google Search results when a query triggers the feature. Not every query will produce one. When it does appear, it offers an AI-generated answer built from information found on the web.
The result can make a search feel more direct. Instead of starting with a list of pages, the user may first see a summary that attempts to answer the question immediately. That can reduce the need to click through to a website or even notice which publications or pages informed the answer.
Google had already offered an early version of the feature inside the Search Generative Experience, which was available only to people who opted in. At the company’s I/O developer conference, Google announced the wider launch of the renamed AI Overviews. Everyone in the US using Google to search the web or ask a question will now see AI Overviews at the top of results when Google’s systems decide a summary is suitable.
You cannot fully turn them off
The most direct answer for users who want old-style search results is simple: there is no complete opt-out for AI Overviews in a Google account. The feature is now part of the default results page.
There are still ways to get closer to a traditional link-focused search page. The manual route is to run a search, then use the results filters. After landing on a page that includes an AI Overview, users can select the More tab and then choose Web. That view shows a results page made up of links to actual websites.
That approach works one search at a time. It does not disable AI Overviews permanently. For people who want the Web view more automatically, browser extensions are another option. The source article notes that developers had already created options for Chrome and Vivaldi, along with add-ons for Firefox, and that other browsers would likely get extensions soon.
What appears inside an AI Overview
The content and layout of an AI Overview can vary. In some cases, links to webpages are visible right away. In others, users need to click Show more before they can see where the information is coming from.
Health-related answers include a disclaimer at the bottom: “This is for informational purposes only. For medical advice or diagnosis, consult a professional.” That warning signals that some answers may need professional context, even when they appear in a search product people use for quick decisions.
Mallory De Leon, a Google spokesperson, described when the feature is expected to appear: “AI Overviews appear for complex queries,” and “You'll find AI Overviews in your Google Search results when our systems determine that generative AI can be especially helpful–for example, when you want to quickly understand information from a range of sources.”
The source article also notes that early testing made the feature feel less predictable, with summaries appearing for simple questions as well as more complex requests. That uncertainty is part of the user experience: people may not know in advance whether a search will produce a standard results page or a generated answer above the links.
How Google says the system is built
According to De Leon, AI Overview is powered by a customized version of Google’s Gemini model, supplemented with parts of Google’s Search system. One example given is the Knowledge Graph, described in the source as having billions of general facts.
Google’s position is that AI Overviews are tied to Search’s existing quality and safety systems. De Leon said: “This implementation of generative AI is rooted in Search’s core quality and safety systems, with built-in guardrails to prevent low-quality or harmful information from surfacing,” and “AI Overviews are designed to highlight information that can be easily verified by the supporting information that we surface.”
That explanation points to the intended safeguard: generated answers should be connected to supporting information. But the experience still depends on whether users open those sources, especially when the summary seems complete enough on its own.
Why verification still matters
The concern is not only that AI Overviews change the look of Search. It is that they place generated text in a position where many people may treat it as the answer.
The source article raises the risk of AI hallucinations, described there as inaccurate generated information. Google’s Gemini chatbot shows a warning that says: “Gemini may display inaccurate info, including about people, so double-check its responses.” AI Overview does not carry that same wording. It often reads, “Generative AI is experimental.”
Google says it used adversarial red-teaming tests to look for weak points in the feature. Still, users who want confidence in a result may need to click through the supporting pages and compare the generated answer with the material behind it.
That extra step may be easy to recommend and hard to expect. Search users often want speed. If an AI Overview appears to answer the question immediately, many people may stop there.
The stakes extend beyond individual searches. Liz Reid, Google’s head of Search, told Lauren Goode that AI Overviews are expected to arrive for countries outside of the United States before the end of 2024, and the feature will likely reach over a billion people. For many of them, this could be a first regular encounter with AI-generated text, even if they have never used a chatbot before.
AI Overviews are meant to make Search faster. The practical challenge is making sure faster answers do not become less checked answers.