We explored London with a little help from Project Astra. ??
With Gemini 2.0, our early research prototype can have more natural conversations, comes with improved memory and can use @Google tools – making it more useful for everyday life. ? pic.twitter.com/3SNSfPaRyn
— Google DeepMind (@GoogleDeepMind) December 11, 2024
Google has unveiled its latest artificial intelligence model, Gemini 2.0, which is set to become the foundation for a range of AI agents designed to perform complex tasks and assist users in various domains. DeepMind CEO Demis Hassabis and Chief Technology Officer Koray Kavukcuoglu highlighted the capabilities of Gemini 2.0-powered agents, which can understand complex instructions, plan and reason, perform actions across websites, and even provide assistance with video game strategies. Several experimental projects are under development, including Deep Research, which helps users explore complex topics by generating multistep research plans and compiling extensive reports, and Project Astra, which aims to function as a universal AI assistant, offering recommendations and advice on various everyday tasks.
Welcome to the world, Gemini 2.0 ✨ our most capable AI model yet.
We're first releasing an experimental version of 2.0 Flash ⚡ It has better performance, new multimodal output, @Google tool use – and paves the way for new agentic experiences. ? https://t.co/ywY2oZv76p pic.twitter.com/1Wgcr3m2Ip
— Google DeepMind (@GoogleDeepMind) December 11, 2024
Project Mariner focuses on creating an AI agent capable of navigating the Chrome browser, while Project Jules is designed as an assistant for developers, integrating directly into a GitHub workflow to aid with coding and planning tasks. Hassabis and Kavukcuoglu emphasized that these projects are still in the early stages of development but are aimed at becoming widely available in future products.
Gemini 2.0-powered AI agents
I'm very proud of the whole Gemini team and everyone else across Google that contributed to all the different announcements today.
Thank you for all the hard work, everyone! ? https://t.co/SESXSq5hB8
— Jeff Dean (@?) (@JeffDean) December 11, 2024
They stated, “We see exciting potential in projects like Mariner, which shows that it’s becoming technically possible to navigate within a browser, though it’s currently not always accurate and slow to complete tasks. This will improve rapidly over time.”
In addition to user assistance, the Gemini 2.0 model has applications in video gaming, offering real-time suggestions based on a wealth of gaming knowledge and collaborating with leading game developers to test these agents. The team is also working on experimental applications of AI agents in the physical world through robotics.
Currently, Google’s AI agents are available only to testers and developers, awaiting further testing and development before broader release. As the AI landscape evolves, industry leaders like Google, Nvidia, and Salesforce are continually exploring and expanding the capabilities and applications of agentic AI across various fields.
