Totally agree with the takeaway here. This never ending cycle of tweaking alignment also is inevitable due to the ever evolving consciousness of humans, which likely also doesn’t have an endpoint. As humans tweak their own preferences and behaviors, value alignment with AI will need to be tweaked as well - both on an individual and societal level. Free speech, expression, and informational exploration will be critical for human well being and growth.
Totally agree with the takeaway here. This never ending cycle of tweaking alignment also is inevitable due to the ever evolving consciousness of humans, which likely also doesn’t have an endpoint. As humans tweak their own preferences and behaviors, value alignment with AI will need to be tweaked as well - both on an individual and societal level. Free speech, expression, and informational exploration will be critical for human well being and growth.
The end of alignment work comes when AI is decisively smarter than human. At that point, it's out of our hands.