Google AI Search Now Sees and Talks About Images

Google has launched a major enhancement to its AI search capabilities, allowing users to combine images and conversational language within the same interaction. This update, now available in English for U.S. users, transforms how people search for items that are difficult to describe using words alone.

The new feature, called visual search, lets you begin with either a text prompt or an uploaded image, then refine the results through natural follow-up questions. According to Robby Stein and Lilian Rincon, Vice Presidents of Product Management at Google, the goal is to solve the common problem of knowing what you want but struggling to phrase it. They explain that now you can simply show or describe your idea to Google and receive a diverse set of visual answers.

For instance, you might start by searching for “maximalist bedroom inspiration,” then ask for “more options with dark tones and bold prints.” Each image displayed will link directly to its original source, making it easy to explore further when something catches your eye.

When it comes to shopping, the experience becomes far more intuitive. Instead of relying on traditional filters for style, size, or color, you can describe what you’re looking for in everyday language. A search for “barrel jeans that aren’t too baggy” will pull up relevant products, and you can further narrow things down by adding “show me ankle length.” This functionality is supported by the Shopping Graph, a massive database containing over 50 billion product listings from major retailers and local stores. Google refreshes more than two billion of these listings every hour to ensure pricing, availability, reviews, and color options are current.

Technically, this visual search upgrade builds upon Google’s existing Lens and Image Search technologies. It incorporates the advanced multimodal and language understanding of the Gemini 2.5 model. A key technique powering this is “visual search fan-out,” where Google automatically runs multiple related queries in the background. This helps the system better interpret the content of an image and the subtleties of a user’s question. On mobile devices, you can even search within a specific picture and ask conversational questions about what appears on screen.

During a recent media briefing, a Google spokesperson provided additional insight. When a search includes subjective terms like “too baggy,” the system may use personalization signals to infer the user’s likely meaning and deliver more tailored results. The spokesperson did not specify which signals are used or how they are prioritized.

Regarding image sources, the system does not currently distinguish between real photographs and AI-generated visuals for this specific feature. However, ranking algorithms tend to favor authoritative and high-quality sources, which can make authentic photos more prominent in search results. Google has not introduced a separate policy or detection standard for AI-generated imagery at this time.

For SEO and ecommerce professionals, this development underscores the growing importance of high-quality imagery. As Google improves its ability to interpret detailed visual and descriptive cues, compelling product photos and lifestyle shots can significantly enhance visibility. Because the Shopping Graph is updated hourly, maintaining accurate and current product feeds is more critical than ever.

As search continues its shift toward visual and conversational inputs, many shopping journeys are likely to start with a simple picture or a casual spoken request rather than a set of precise keywords. This new visual search experience is currently rolling out in the U.S. in English, with no official timeline yet announced for expansion to other languages or regions.

(Source: Search Engine Journal)