Artificial Intelligence A Review Paper

AI research has focused on improving the decision-making capabilities of computers, i.e., the ability to pick high-quality actions in pursuit of a given objective. When the aim is aligned with the values of the humanity, this will cause tremendous benefits. When the target is misaligned, improving the AI system’s decision-making may cause worse outcomes for the humanity. The objectives of the proposed research are to make a mathematical framework during which fundamental questions useful Alignment is often investigated; to develop and experiment with methods for aligning the values of a machine (whether explicitly or implicitly represented) with those of humans; to know the relationships among the degree useful alignment, the decision-making capability of the machine, and therefore the potential loss to the human; and to understand especially the implications of the computational limitations of humans and machines for value alignment. The core of our technical approach are going to be a cooperative, game-theoretic extension of inverse reinforcement learning, allowing the various action spaces of humans and machines and therefore the varying motivations of humans; the concepts of rational meta-reasoning and bounded optimality will inform our investigation of the effects of computational limitations.