Google has announced major improvements to its search engine, introducing a new experimental feature called AI Mode, which uses the upgraded Gemini 2.0 language model, and expanding the capabilities and availability of AI Overviews. These updates are designed to give users more detailed and advanced search results.
AI Overviews, which more than a billion people already use, are getting a significant upgrade with the integration of Gemini 2.0 in the U.S. This update allows AI Overviews to handle more complicated questions, starting with topics like coding, advanced math, and queries that involve multiple types of information.
Google says that Gemini 2.0 provides faster and better-quality answers, which means AI Overviews will appear more often for these types of questions. Additionally, more people can now use AI Overviews; teenagers have access to the feature, and users no longer need to sign in to use it.
The main highlight of Google’s announcement is the launch of AI Mode, an experimental feature available through Search Labs. AI Mode is designed to meet the growing demand for AI-powered answers across a wider variety of searches. It builds on the foundation of AI Overviews but adds more advanced reasoning, thinking, and the ability to handle multiple types of information.
This allows users to ask complex, multi-part questions and follow up with additional queries to explore topics in greater depth. AI Mode uses a customized version of Gemini 2.0 and is particularly good at answering questions that require exploration, comparisons, and detailed reasoning. It can handle nuanced queries that would usually require several searches, such as comparing detailed options or exploring new ideas. The responses also include links to relevant websites for further reading.

One of the key technologies behind AI Mode is called “query fan-out.” This method simultaneously runs multiple related searches across different subtopics and data sources, including the Knowledge Graph, real-world information, and shopping data for billions of products. The results are then combined into a single, easy-to-understand response, offering more depth and breadth than a traditional Google search.
For example, if someone asks about the sleep-tracking features of different devices, AI Mode would perform a multi-step process that includes planning, gathering information, and adjusting the plan based on what it finds.
At first, access to AI Mode will be limited. Google is starting with Google One AI Premium subscribers, giving them early access through Search Labs. Users can try AI Mode by visiting Search Labs, going to google.com/aimode, or using a special icon in the Google app on mobile devices.
Google emphasizes that AI Mode is still experimental. The company is committed to improving the user experience and expanding the feature’s capabilities based on feedback. Future updates may include more visual responses with images and videos, better formatting, easier access to web content, and addressing potential challenges like ensuring objectivity in responses and avoiding unintended biases. Google also notes that if AI Mode isn’t confident in its answer, it will show standard web search results instead.
Published: Mar 5, 2025 05:30 pm