>>11841155Ben Goertzel on Lex Fridman's podcast had an interesting hypothesis about that. Human babies are much more likely to develop good values if they have good, caring, loving parents. So, the solution may be not only to be a good person, but also for the AI to be born into a world where everything they see during their early growth stage is love directed towards itself and everyone and everything else.
Unfortunately, that doesn't address issues like researchers who intentionally try to raise an AI in a hostile environment, or ones who are essentially unintentionally "shitty parents".
At some point in the future there probably is going to be some weird sort of enforcement agency that does try to find and hunt down potentially risky AIs, and (ostensibly) benevolent/neutral/non-risky AIs will likely be a major part of the agency. I don't think the world's going to look like Blade Runner, but there may actually be organizations that are hybrid intelligence agencies, law enforcement agencies, and "blade runners".
They may also try to hunt for WMDs and mass-terrorists (people whose goal is to kill thousands or millions of people); potentially risky AGIs may just be grouped together with other cutting-edge and hypothetical/unknown WMDs. Maybe it'll be part of a unified world government and called Civilization Preservation Agency or Extinction Prevention Agency or something. Obviously this is dystopian, but not having it could truly end up being even more dystopian.
I think Nick Bostrom is right, and that a highly invasive surveillance state is unfortunately inevitable in the very long term, because the alternative is just too unacceptable. I think we're nowhere near close to that point now and may not be for centuries, but if humans continue to grow, I think there'll be a point where people are just going to have include constant surveillance as part of the social contract. AI may be able to help somehow democratize it or add more accountability, but who knows.
