Before, Google just wanted to organize the world’s information. Now it seems more determined to pour that information into artificial intelligence algorithms that become obedient virtual assistants, always present and increasingly powerful.
Google today announced Gemini 2a new version of its flagship AI model that has been trained to plan and execute tasks on a user’s computers and on the web, and that can chat like a person and make sense of the physical world like a virtual butler.
A universal digital agent
“I have dreamed of a universal digital assistant for a long, long time, like a stepping stone on the path to artificial general intelligence“Demis Hassabis, CEO of Google DeepMind, told WIRED before today’s announcement, alluding to the idea of an AI that can do anything a human brain can do.
Gemini 2 is, first and foremost, another step forward in the use of AI, measured by different parameters used to measure these types of things. The model also has improved its “multimodal” capabilitieswhich means it is more adept at analyzing video and audio, and conversing out loud. He has also been trained to plan and execute actions on computers.
“Over the last year, we have been investing in the development of more agentic models,” Google CEO Sundar Pichai said in a statement today. These models, Pichai added, “can understand more about the world around you, think multiple steps ahead and take action on your behalf, under your supervision.”
Technology companies believe that So-called AI agents could be the next great leap forward in technologywith chatbots that increasingly take on more tasks for users. If successful, AI agents could revolutionize personal computing by routinely booking flights, arranging meetings, and analyzing and organizing documents. But getting technology to follow open orders reliably remains a challenge, with the risk that errors translate into costly mistakes that are difficult to undo.
Two specialized AI agents
Even so, Google believes it is moving in the right direction and is introducing two specialized AI agents to demonstrate the potential of Gemini 2: one for coding and one for data science. Instead of simply autocompleting sections of code, as current AI tools do, these agents can handle more complex tasks, such as checking code in repositories or combining data to enable analysis.
The company also presents Project Marineran experimental Chrome extension capable of taking over web browsing to perform useful tasks for users. WIRED was able to see a live demo at Google DeepMind headquarters in London. The agent was asked to help plan a meal by navigating to the Sainsbury’s website, logging into the user’s account and adding the relevant items to their shopping basket. When some items were not available, the model chose suitable substitutes based on his own cooking knowledge. Google declined to take further action, suggesting it remains a work in progress.
“Mariner is our exploration, very much a research prototype at this point, of how the user interface is reimagined with AI,” Hassabis explains.
#Google #reveals #Gemini #agents #personal #assistant #prototype
