Nate Silver
I love building empirically grounded statistical models. They work well in data-rich environments like sports. AI x-risk isn't like that. Few technologies are good analogies. Plus, there's survivorship bias, and the track record on nuclear weapons isn't super reassuring, btw.