Google is pushing forward with a bold transformation of its Search product, aiming to make it function more like an AI assistant instead of a traditional search engine. During the company's latest earnings call, CEO Sundar Pichai explained how AI will play an even larger role in Search, citing 2025 as a watershed moment for progress in this sector.
This shift began in 2024, when AI Overviews, a feature highlighting responses at the top of search results, was introduced. However, Google is now trying to expand beyond simple search summaries by using powerful AI models like Project Astra and Gemini Deep Research.
Astra, a multimodal AI system developed by Google's DeepMind, can process live video input, evaluate what it sees, and reply in real time. Meanwhile, Gemini Deep Research can develop long-form research reports by scouring the web, automating tasks traditionally performed by users.
Another significant advancement is Project Mariner, an AI agent that can interact with websites on people's behalf. If implemented, this could reduce the need for people to visit webpages on their own, creating concerns among businesses and content creators that rely on Google-driven traffic. Additionally, Pichai stated that future Search experiences may include more chatbot-like interactions, allowing users to ask follow-up questions for a more conversational feel.
This AI-driven innovation comes as Google faces more competition from AI-first search engines like ChatGPT. While Google Search still dominates with 16.5 billion U.S. visitors in December 2024—30 times higher than ChatGPT—OpenAI’s tool is rapidly gaining traction. Google is responding by combining AI capabilities with its existing search infrastructure to stay competitive.
However, Google's artificial intelligence journey has not been without obstacles. When AI Overviews were launched, they delivered bizarre and inaccurate results, such as advising users to eat rocks or put glue on pizza. Despite these early setbacks, Google continues to work on transforming Search into an AI-powered tool that actively evaluates information for users. Whether this transformation will succeed or further complicate things for users and businesses remains to be seen.