Google Expands AI Mode Access in Search Labs

Google has expanded access to its "AI Mode" search experiment to millions of users enrolled in Google Labs, moving the feature beyond its initial restriction to Google One AI Premium subscribers [3]. The tool, which allows users to ask complex, multi-part questions and perform follow-up inquiries, represents a direct challenge to AI-native search services like Perplexity and OpenAI’s ChatGPT Search [4].

Expanding the experimental search footprint

AI Mode utilizes a technique called "query fan-out," which issues multiple related searches concurrently across various data sources to synthesize a comprehensive response [4]. The system is designed to handle queries that require advanced reasoning and multi-step exploration, such as comparing product features or identifying items in uploaded photos [3]. According to Google, users testing the feature are submitting queries roughly twice the length of traditional search inputs and are utilizing follow-up questions in approximately one-quarter of their sessions [4].

The latest update adds multimodal capabilities, allowing the model to analyze the materials, colors, and spatial relationships of objects within user-uploaded images [3]. While the model is trained to prioritize factuality and cite sources, Google acknowledges the experimental nature of the technology and notes that the system may still produce errors [4]. In instances where the model lacks confidence in its output, it is programmed to provide a list of relevant web links rather than attempting an AI-generated summary [4].

Competitive positioning

Google’s push into AI-driven search comes as the company seeks to maintain its market position against competitors that have gained traction with conversational search interfaces [4]. By integrating AI Mode into Google Labs, the company is leveraging its existing infrastructure—including the Knowledge Graph and real-time shopping data—to differentiate its offering from standalone AI search startups [4].

The company is currently iterating on the user experience, with plans to incorporate more visual elements and surface content from a broader range of sources, including user-generated material [4]. This expansion follows the recent deployment of Gemini 2.0 within AI Overviews, which now supports advanced tasks like coding and complex mathematical problem-solving [4].

What to watch

• Feature Rollout: Monitor whether Google transitions AI Mode from the experimental Labs environment to a permanent, standard feature within the main Google Search interface.
• Multimodal Integration: Watch for further updates regarding the model's ability to process and act upon diverse media types, such as video-based "how-to" queries.
• Competitive Response: Observe how rivals like Perplexity and OpenAI adjust their own query-length and reasoning capabilities in response to Google’s "query fan-out" technique.

The success of AI Mode will likely depend on whether Google can balance the utility of its advanced reasoning capabilities with the reliability required for high-stakes information categories like health and finance. As the company continues to refine the model's ability to cite sources and prioritize accuracy, the primary question remains how quickly these experimental tools can be scaled to the entirety of the global search user base.