Maybe AI systems would be safer if they avoid gaining too much control over their environment? How might that work?This is a follow-up to this earlier video:...