It's bad for there to be anything near us that exceeds our (collective) cognitive capabilities unless the human-capability-exceeding thing cares about us, and n

It's bad for there to be anything near us that exceeds our (collective) cognitive capabilities unless the human-capability-exceeding thing cares about us, and no one has a good plan for arranging for an AI to care about us even a tiny bit. There are many plans, but most of them are hare-brained and none of them are good or even acceptable. Also: no one knows with any reliability how to tell whether the next big training run will produce an AI that exceeds our cognitive capabilities, so the big training runs should stop now.