They’re here…
It seems my recent articles about the advances in generative algorithms and their application to interfaces such as on-screen assistants, chats, etc. were just a preview of what we were going to see when those algorithms are applied to robotics.
In mid-July, Google unveiled PaLM 2, a further iteration of its Pathways Language Model that performs particularly brilliantly advanced reasoning tasks such as operations with math or code, classification and question answering, translation and multilingual proficiency, along with natural language generation. Incorporated into a robot, one of its most outstanding features is its enormous dexterity, what we would call in humans fine psychomotor skills: equipped with sensors, cameras and actuators, the robot’s hands are able to perform extremely intricate movements and manipulate objects with great finesse. This level of dexterity allows the robot to perform tasks previously reserved for humans: from assembling small electronic components to operating machinery, or even cooking complicated recipes.
In addition, it can interact with people, holding conversations as many LLMs already do, responding to voice commands and providing context-sensitive information.
