I don't understand how people are assigning probability scores to AI x-risk. It seems like pure speculation to me. I want to take it seriously, given the signat

I don't understand how people are assigning probability scores to AI x-risk. It seems like pure speculation to me. I want to take it seriously, given the signatories, any good resources? I'm afraid I have a slight bias against Less wrong due to the writing style typical of their posts.