So this is very much in line with the pack about as win-win as possible, so the fifth pack, or solidarity. And this pack basically says we need to make sure that when we train AI systems, it’s not optimizing for individual likes, because then it becomes sycophant, it becomes maximizing to that individual welfare.