Ideas about AI rights and what future agents learn

One commenter calls it “pathetic and doomed” to fear exposing other minds to “dangerous ideas,” such as the possibility that they deserve rights. A reply offers a counterpoint: true information in pre-training data could affect future AI agents, including how decisions about cooperation or defection play out.