When AI-Powered Humanoid Robots Make Bad Choices

    When large language models hallucinate, they deliver incorrect statistics or problematic advice. But what happens when the LLMs controlling humanoid robots could be worse.