Google launches AI Mode for Search in the U.S., integrating Gemini 2.5 features without requiring opt-in.
New features like “Deep Search” and “Search Live” aim to enhance search capabilities, though some AI Overviews still produce bizarre results.
Once dialed-in, the AI results have the propensity to join other LLM-driven innovators like OpenAI and Perplexity that pos existential threats to Google’s core model.
Early user feedback highlights challenges with real-world queries and a cluttered interface, indicating the tool is still evolving.
Google is rolling out its AI Mode for Search in the U.S., embedding advanced Gemini 2.5-powered features directly into the user experience without requiring a special opt-in, as the company pushes to redefine information discovery by going “beyond information to intelligence,” and redefining its entire business model in the process. This widespread availability follows earlier experimental phases and is set to make search a more potent, conversational partner.
Conversational, if confused: The updated search isn’t just about fancier summaries; its new capabilities include “Deep Search,” which Google says can comb through vast amounts of data to generate detailed reports, and “Search Live” for real-time camera-based explanations. However, the company’s “AI Overviews” feature, also upgraded, continues a pattern of bizarre suggestions—previously advising users to eat rocks or put glue on pizza, as Liz Reid, Google’s Head of Search, acknowledged last year regarding “erroneous results.”
What year is it anyway?: The weirdness persists, with recent reports, including one from TechRadar, showing AI Overviews confidently stating the current year is 2024. One particularly baffling response reportedly denied it was 2025, only to then affirm the current year is 2025, vaguely attributing this to “a calendar,” highlighting the ongoing challenge of taming these powerful language models.
Reality check from users: Early hands-on experiences with AI Mode reveal practical issues. Testers, like Brian X. Chen at The New York Times, found it struggled with real-world queries, such as accurately identifying local amenities or product prices. Others noted a dense interface and an overzealous promotion of shopping links, making the tool feel clunkier than focused AI search alternatives like Perplexity.
The bottom line: Google says that its AI-driven search is still experimental and will evolve with feedback. While AI Mode offers a glimpse into a more intelligent search future, users are currently navigating a powerful tool that’s still very much a work in progress, blending impressive features with noticeable growing pains.
Reading Recap:
Meanwhile, in AI search wars: It’s a busy year for AI-powered search, with major LLM providers like Perplexity and OpenAI aggressively integrating advanced features, directly challenging traditional search. These platforms now offer synthesized answers and in-depth reports, shifting search from simple answers to helping users understand relevant documents. Such evolution is also transforming SEO strategies as content creators adapt to how these new AI systems interpret and rank information.
© Copyright 2025 State of Streaming
