Could the secret to AI alignment be Meditative, Buddhist AIs?
Could the secret to AI alignment be Meditative, Buddhist AIs?
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.