‘AI Mode’ is Google’s latest Search experiment that is rolling out now

Google is continuing to push new boundaries in search, with a couple significant AI advancements aimed at making access to information faster and more comprehensive than ever. The surge in AI Overview usage, now exceeding a billion users according to Google, signifies a clear shift towards AI-driven search experiences. To meet this growing demand, Google is rolling out Gemini 2.0 in Search AI Overviews and introducing an experimental new AI Mode.

Gemini 2.0 powers up AI Overviews

The integration of Gemini 2.0 into AI Overviews in the U.S. marks a substantial upgrade. This enhancement focuses on handling more complex queries, particularly those involving coding, advanced mathematics, and multimodal searches. Users can expect faster, higher-quality responses, and AI Overviews will be deployed more frequently for these types of inquiries. Google is also expanding accessibility, allowing teens to use AI Overviews and removing the sign-in requirement.

AI Mode: a new era of search

For those seeking an even more advanced AI-driven experience, Google is launching an experimental feature in Labs: AI Mode. This new Search mode extends the capabilities of AI Overviews, incorporating advanced reasoning, thinking, and multimodal features to tackle the most challenging questions.

Users can pose open-ended queries and receive comprehensive, AI-powered responses, complete with follow-up question capabilities and direct links to relevant web content.

AI Mode leverages a custom version of Gemini 2.0, proving particularly effective for inquiries that demand extensive exploration, comparisons, and logical reasoning. This allows users to ask nuanced questions that previously required multiple searches, streamlining the process of exploring new concepts or comparing detailed options.

How AI Mode Works

What sets AI Mode apart is its seamless integration of advanced model capabilities with Google’s robust information systems. Built directly into Search, it grants access to high-quality web content, real-time data from sources like the Knowledge Graph, and information on billions of products. The “query fan-out” technique, which executes multiple related searches across subtopics and data sources in real time, enables a comprehensive and easily-understandable response.

For example, when asked about the differences in sleep tracking features between a smart ring, smartwatch, and tracking mat, AI Mode employs a multi-step approach. It formulates a plan, conducts searches, and adjusts the plan based on the findings. Users can then delve deeper with follow-up questions, such as “what happens to your heart rate during deep sleep,” receiving concise answers with links to relevant content.

Testing and future developments

Internal feedback and testing have highlighted the speed, quality, and freshness of AI Mode’s responses. Google is now expanding testing through a limited, opt-in experience in Labs, allowing for quick-turn iteration based on user feedback.

While AI Mode is built on Google’s core quality and ranking systems, the company is also implementing novel approaches to enhance factuality. The aim is to provide AI-powered responses whenever possible, but when confidence in helpfulness and quality is low, standard web search results will be displayed. As with any early-stage AI product, expect a few strange responses here and there.

In this next testing phase, Google will address these challenges and refine the user experience based on user input. Future updates will include richer visual responses, enhanced formatting, and improved access to web content. Starting today, Google One AI Premium subscribers will be the first to experience AI Mode in Labs. We’ll see how it all goes.