Source: Sex and the State
by Cathy Reisenwitz
“The issue of AI safety does not appear to me to be an instance of a problem where the answer is obvious but hard-to-execute. AI safety seems like a values problem, to me. And those disagreements are always the thorniest. There is no way to develop totally safe AI. When you build a bridge or road, you build it knowing that someone is going to die on it. Probably a lot of someones. But you build it anyway, because you think the benefit is worth the cost. We have to think of AI this way. The people building AI are thinking of it this way. The goal cannot be ‘no one dies.’ The goal has to be, ‘Here’s who can die and under what circumstances, and how often, and with what recompense, and for what benefit.'” (03/19/26)
https://cathyreisenwitz.substack.com/p/no-good-options-for-ai-safety