This is a paper which argues that the true path to a safe, dependable AI system is to take what we’ve learned from meditation and Buddhism and apply it to AI systems: “Robust alignment strategies need to focus on developing an intrinsic, self-reflective adaptability that is constitutively embedded within the system’s world model, rather than using brittle top-down rules”, the authors write.
You must log in or register to comment.
Source: Trust me, I’m a philosopher.
It’s pretty light on data, and pretty heavy on supposition. Also TIL Imperial College has a “Centre for Psychedelic Research”. This is one of the ideas they got while “researching”, I guess?
What ever it is you’re smoking, stop smoking it.
