Tag
This paper introduces Constructive Alignment, a paradigm that reframes AI alignment as governing the evolution of human preferences over time rather than satisfying static preferences. It proposes a control-theoretic framework to regulate how AI systems influence value trajectories.
This paper introduces a multi-agent environment based on the board game Fog of Love to evaluate affinity-based reinforcement learning for instilling virtuous behavior in AI agents. The authors demonstrate that localized affinities improve agent performance in both competitive and cooperative objectives, advancing machine ethics research beyond simple grid-world environments.