Google announced Gemini 2, a major update to its flagship AI model, designed to upend personal computing, web search, and how people interact with the physical world. The model boasts advanced “agentic” capabilities to plan and execute tasks, converse like a human, and process audio and video content.
Gemini 2 is supposed to be able to do tasks like booking flights, scheduling meetings, and document management. Google also introduced two specialized AI agents for coding and data science that could go beyond simple task automation to complex workflows, like managing repositories and doing data analysis.
One of the highlights is Project Mariner, an experimental Chrome extension that enables AI to navigate websites and execute helpful tasks, such as grocery shopping. During a demo, Mariner signed into a user’s supermarket account, added items to the cart, and even selected replacements when products weren’t in stock showcasing how it can adapt and make decisions independently.
Another feature, Astra, extends the capabilities of Gemini 2 into the physical world. Using a smartphone camera, Astra can analyze surroundings for context, and converse with people naturally. For instance, it identified wine bottles, gave tasting notes, and sourced pricing from the web. It also offered detailed insights about paintings and instantly translated text during a live demonstration.
Google DeepMind CEO Demis Hassabis positioned Gemini 2 as a potential universal digital assistant and a step toward artificial general intelligence. “This could transform how people interact with technology,” Hassabis said, highlighting its learning capabilities and user customization options.
The announcement signals Google’s renewed push to compete with OpenAI’s ChatGPT, following the initial launch of Gemini in December 2023. While the technology holds transformative potential, Hassabis acknowledged the challenges of ensuring privacy and preventing misuse, stating, “We need to think about security and privacy very seriously upfront.”
Although Gemini 2 is still in development, its features underpin Google’s ambition to redefine the role that AI plays in everyday life while offering glimpses into a potential future where AI assistants will be sewn into personal and professional contexts.
