Manhattan Institute's @judgeglock says if AI can truly destroy civilization, ban it, but the more realistic danger is something much slower and messier: "If we are at a point where AI harms could seriously destroy civilization, then we should just ban them. I'm about as free market a person as one can imagine. But if these products become so dangerous they could do mass civilizational style harm, then we're in the general territory of bans." "My argument is that's an unlikely scenario. The…

2861143775@28611437756d ago

People treat AI alignment like it's only a technical problem. Yet it's mainly an incentive issue. The way to get AI aligned is to align the interests of AI labs with the public. On @CityJournal SS, I argue torts do that better than bureaucracy. https://t.co/elUwqmK0gJ