Stanford robotics prof. Allison Okamura warns that putting ChatGPT-style AI inside robots could turn hallucinations into real-world injuries: "Generalist robot bodies are especially exciting because you could use generative and agentic AI in new ways. But the controversy is that it’s really hard to have safety guarantees." "We’ve all experienced hallucinations from LLMs, and with robots, the danger is much higher, because you don’t know what the robot is going to do until it makes that…