Google DeepMind, a subsidiary of tech giant Google dedicated to developing AI technology, has recently released a paper on RoboCat. This self-improving, AI-powered tool can perform multiple tasks.
In a blog released on their website, DeepMind explains the potential for this latest AI agent: “Our latest paper introduces a self-improving AI agent for robotics, RoboCat, that learns to perform a variety of tasks across different arms, and then self-generates new training data to improve its technique.”.
The key feature of RoboCat is its ability to perform new tasks through a process of self-learning, which has caught the attention of both the AI and robotics communities. Whereas traditional AI uses training data (data used to teach an AI model), Google’s RoboCat can also generate its data and implement both kinds of data to perform newer tasks.
This self-generation of data creates a rigorous training cycle where the more activities Robocat learns, the better it becomes at solving additional tasks. Comparing this learning curve to how humans learn from experience, DeepMind states, “RoboCat’s ability to independently learn skills and rapidly self-improve, especially when applied to different robotic devices, will help pave the way toward a new generation of more helpful, general-purpose robotic agents.”.