Google DeepMind Unveils AI Models That Enable Robots to Plan Complex Tasks and Search Internet Independently

Google DeepMind releases two groundbreaking AI models that transform robots into internet-savvy problem solvers capable of planning complex multi-step tasks like sorting laundry by color and packing weather-appropriate suitcases, achieving 20-40% success rates in early artificial general intelligence demonstrations.

• Google DeepMind releases two new AI models that enable robots to plan multi-step tasks, search the internet for information, and adapt to new situations without specific programming
• The Gemini Robotics 1.5 and Gemini Robotics-ER 1.5 models allow robots to perform complex tasks like sorting laundry by color, packing suitcases based on weather forecasts, and checking recycling rules online
• The robots demonstrate early artificial general intelligence capabilities with success rates of 20-40% in complex tasks, marking a shift from single-task machines to adaptable general-purpose robots