Embedding Morality in Artificial Agents through Reinforcement Learning and Explicit Principles

Combining top-down moral principles with bottom-up reinforcement learning is a promising approach for developing safe and adaptable moral AI agents.