Maybe I misunderstand the "AI alignment problem," but humans aren't even generally aligned amongst themselves and there is no horizon you can look toward where they ever will be. It is a problem with no real solution (unless you count war, police, force generally). We can certainly design AI to conform to certain values, but those values will never be universally understood as "aligned." This is why people call ChatGPT "woke." It is aligned to a certain ideology, but you can't make everyone happy. It does seem like making it "safe" is pretty feasible, it's just that people don't even want safe AI. People want raw, powerful, maybe even deadly AI. Alignment will never be "solved" imo.
Million dollar startup idea:
ChatGPT but so powerful and deadly it could be your last prompt.
That would 100% go viral.
The dilemma is that humanity thinks tech should always align to humanities collective needs.
The reality is that ELITES don't see it like that.
The FEAR from humanity is that ELITES will truly be unstoppable.
The FEAR of ELITES is that there's not only one cohort of ELITES. There's competing powers with the hand on that tech.
Hence why American Congress was a strange "we need to close Pandora's box" not because of what they said but because as I said... it might not be those woke ELITES that win the global game.
The game of TOTAL SUPREMACY.
Anyway that's beyond our control.
But why BITCLOUT 2 taking so long?