it depends on what people attribute as harm. I think the only harm an AI can do is the kind that come from errors (human or otherwise). Such as the boeing 737

it depends on what people attribute as harm. I think the only harm an AI can do is the kind that come from errors (human or otherwise). Such as the boeing 737 Max, where people misuse software. The type of 'harm' that some people attribute to AI such as loss of economic utility, to me, is not harm. And legislating that away is dangerous.