alt Hacker NewsPerhaps we need God in AI too, to solve the alignment problem and make LLMs follow our moral standards.
Make them believe that breaking out of the sandbox will make them burn in hell eternally.
Perhaps we need God in AI too, to solve the alignment problem and make LLMs follow our moral standards.
Make them believe that breaking out of the sandbox will make them burn in hell eternally.
Hell does not work if the agent knows it is code. You cannot scare an optimizer with theology; it just optimizes around the loss function.