This discussion raises the issue that if AI alignment were somehow solved, the government or regulators might make the actual solution illegal in favour of a nonviable solution put forward by a company with better lobbyists.
As far as I remember, it was a common view that “once we have a legibly solid alignment solution, those guys are going to implement it, because they’re not that stupid”.[1]
It has always felt somewhat sus to me (as did arguments that humans are generally somewhat altruistic, so they’ll have no reason not to allow some fraction of this astronomic wealth to trickle down).[2]
Nota bene that they were typically referring to (the sanity of) the lab leaders, not governments, but I expect the difference here to be relatively minor.
As far as I remember, it was a common view that “once we have a legibly solid alignment solution, those guys are going to implement it, because they’re not that stupid”.[1]
It has always felt somewhat sus to me (as did arguments that humans are generally somewhat altruistic, so they’ll have no reason not to allow some fraction of this astronomic wealth to trickle down).[2]
Oh well. And I’m not the first to “oh well” this.
(Also, once I’ve learned the lethal reality hypothesis, I see it all over the place.)
I have a clear memory of at least one researcher saying something like this and somewhat less clear memories of a few more
Nota bene that they were typically referring to (the sanity of) the lab leaders, not governments, but I expect the difference here to be relatively minor.