For the past two decades, Google has relied on algorithms to deliver relevant web links for user search queries. However, at the I/O 2025 event, the company indicated a significant shift away from traditional search methods. Google CEO Sundar Pichai and his team revealed innovative approaches for users to explore the web through AI agents.
Liz Reid, Google’s VP of Search, expressed excitement about the new chapter in search capabilities where users can pose any question, from the simplest to complex research needs and personalised shopping queries. Google’s latest update introduces an AI mode for all U.S. Search users, allowing them to interact with an AI agent that can browse websites, summarise content, or aid in shopping. Additionally, Project Mariner will offer Ultra subscribers an advanced AI agent capable of multitasking, visiting and interacting with multiple web pages simultaneously.
Google’s Deep Research agent also received enhancements, enabling it to generate detailed research reports while integrating with users’ Gmail and Drive. The company is further enhancing Project Astra—its multimodal AI experience—by embedding it into Search and Gemini, facilitating more conversational interactions and visual inputs from users.
The emergence of ChatGPT has propelled Google to rethink its information retrieval methods, a trend that started at last year’s I/O with the introduction of AI overviews, despite earlier technical hiccups. The firm’s latest developments at I/O 2025 indicate a robust strategy where AI agents could transform how users access and digest information online.
Google announced that its SDK for the Gemini models will now support Anthropic’s MCP, a growing standard connecting AI agents to internet data sources. This move aligns with a broader industry trend, as other tech leaders, such as Microsoft’s CTO, advocate for an “open agentic web,” emphasising seamless communication between agents and data sources.
Despite the optimism surrounding AI’s role in reshaping the internet, challenges remain. Critics point out that deploying AI agents to websites threatens the ad-supported model, which could adversely impact industries relying on advertising revenue, particularly publishers who may struggle to compete for audience attention against AI-generated summaries.
During I/O, a Google representative highlighted the urgency of reclaiming user time with AI agents, but concerns persist regarding the accuracy of AI outputs. Instances of AI “hallucinations”—misinformation generated by algorithms—remain problematic and could undermine user trust in online information.
The implications of these changes could be profound. Should AI agents dominate web interactions, scepticism towards information reliability may rise, compounding the challenge of misinformation. However, Google appears committed to advancing its AI initiatives, indicating a significant pivot in their vision for the future of the web, centred around AI agents. This transformation may alter the landscape of online information consumption and content creation dramatically.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
