China and America race to develop robots that learn new tasks without prior programming

China and the United States are competing to develop robots that can learn new tasks without prior programming, using artificial intelligence systems that enable them to understand their surroundings and adapt to unfamiliar situations. Chinese and American startups are working on similar technologies, and 'Generalist' announced that its model 'Gen 1.5' completed 60% of simple tasks after watching a demonstration.

China and the United States are racing to develop robots capable of learning new tasks without prior programming, using artificial intelligence systems that allow machines to understand their environment and adapt to unfamiliar situations. This approach focuses on developing software that makes robots more autonomous, rather than relying on pre-programmed instructions for each task. Yang Zhang, co-founder of Chinese company 'X Square', said the goal is to move robots from relying on pre-defined software to a stage where they can act based on something like common sense. This shift relies on AI models that analyze what robots see, understand relationships between objects, and learn how to interact with them. American startups such as 'Physical Intelligence', 'Skilled AI', and 'Generalist' are working on similar technologies. 'Generalist' announced that its model 'Gen 1.5' managed to complete about 60% of simple tasks after watching a demonstration.