The question of whether AI systems are conscious or sentient is largely irrelevant to AI risk because what matters is what AI systems do and are capable of doing—their functional capacities create risks regardless of whether anyone experiences anything