How safe is our reliance on AI, and should we regulate it?

Some of the writing that GPT-3 produced has been amazingly good; but some has also been racist, sexist, or otherwise perfidious text, due to the built-in biases of the data that it mines to teach itself to write.