The OpenAI tussle is between the faction who think Skynet will kill them if they build it, and the faction who think Roko's Basilisk will torture them if they don't build it hard enough.
@jef This caused me to read about Roko's Basilisk and wow, that entire concept is just beyond mockery. These people think they're doing some sort of rational analysis?

@KindlyWizard @jef

It's an agnostic version of Pascals wager, which in fact was taken seriously (and which also arose in other belief systems even before christianity)

@phoyd Thanks for explaining PHI 101 to me.

If you swap out "Gargamel, the evil wizard from the Smurfs" for "AI" you get the same "results".

@KindlyWizard nah, I think the Basilisk needs to be God-Like to make this work with the eternal punishment and so one. We need imagine something that can travel back in time and hurt us.
@phoyd maybe _you_ need to. I'm good, thanks.
@KindlyWizard I dont need this too. The whole argument boils down to: if there is a nonzero probability of event X with payout -inf, then my average payout is -inf, which apparently gives some people an existential crisis. Otoh we could all be Boltzman brains and have no reason to bother at all.