I find it very ironic that the allegedly rational community of lesswrong chose to pascal-wager themselves so hard…

They’re rational in the sense of following utilitarianism to its endpoint. (This is why Pascal’s wager poses such a problem.)

It would be even more ironic if they’re committing the same error they fear AIs would: namely, to take a simple defined goal function or system and be led astray by taking it to its ultimate conclusion.

What if someone else is faster and they build a different Basilisk that will have different values and will torture those who participate in creating any Basilisk that uses acausal blackmail?
The people who create this Basilisk may do so because the prospect of this Basilisk being realized can be used as a way to prevent people from being afraid of the original Basilisk.
You see, you can imagine all sorts of Basilisks. Even irrational ones that don’t respond to reasoning. Ones that torture everyone, for example.