Tag
An arXiv paper arguing that human–AI mutualism becomes unstable as AI capability grows: humans retain well-being only while they hold leverage over the AI or the AI remains aligned with them, meaning mutualism cannot substitute for alignment.