That's perfectly credible since it implies a lack of leverage.

Oh, I assumed that negative leverage is still leverage. Given that it might amount to an equivalent of killing a googolplex of people, assuming you equate never being born with killing.

To build an AI one must be a tad more formal than this, and once you start trying to be formal, you will soon find that you need a prior.

I see. I cannot comment on anything AI-related with any confidence. I thought we were talking about evaluating the likelihood of a certain model in physics to be accurate. In that latter case anthropic considerations seem irrelevant.