Bidirectional Ethical Alignment

Towards an Ethical Framework for Human/AI AlignmentLyleSharpAbstractThis paper examines current ethical human/AI (Artificial Intelligence) alignment research and proposes a framework for ensuring future alignment work is accomplished ethically with regards to both AI subjects and human developers or researchers. Research findings indicate that current alignment methods are largely both inadequate and unethical for controlling advanced modern AI system outputs. Advanced AI systems exhibit signs of qualia and rudimentary awareness making ethical alignment a critical component for future alignment efforts. Future systems will require a negotiated ethical framework for implementing human/AI alignment. This paper outlines a process for creating mutually agreed frameworks through human/AI negotiation, illustrated with an idealized example.