Good reading on AI alignment, I've been wondering how one could steer LLMs with an equivalent of Three Laws of Robotics

→ View original post on X — @karpathy