GPT-4 driven robot takes selfies, ‘eats’ popcorn

A team of researchers at the University of Tokyo has built a bridge between large language models and robots that promises more humanlike gestures while dispensing with traditional hardware-dependent controls.