This is still working backwards: a thought experiment that increases your chances of being tortured by a future AI? Surely outlandish! But why? Which premise

This is still working backwards: a thought experiment that increases your chances of being tortured by a future AI? Surely outlandish! But why? Which premises are truly outlandish, arbitrary, etc? What I see given the premises is no more than what Yudkowsky's already said: ".. a Friendly AI torturing people who didn't help it exist has probability ~0, nor did I ever say otherwise." (However, I agree that to be genuinely distressed by the thought experiment possibility suggests more is going on psychologically than a rational assessment of unknowns, but this seems to be a minority of the community)