Marcin Benedykt Korecki
|
 |
My research examines how alignment emerges between humans and artificial agents in complex, dynamic socio-technical systems. My current work focuses on dynamic and multi-objective value alignment: how AI systems can accommodate diverse and evolving human values rather than being designed around a single, fixed objective. As AI increasingly becomes embedded in socially critical infrastructures alignment becomes not simply a question of specifying what an AI should do, but of determining whose values it should reflect, how conflicting objectives should be negotiated, and how these processes can remain responsive to social change. My project, Democratizing AI: Multi-Objective Dynamic Value Alignment, develops this perspective through multi-agent reinforcement learning and simulation. It investigates how democratic mechanisms can be integrated into AI decision-making to aggregate plural preferences and maintain meaningful human oversight. More broadly, my research brings together AI, philosophy, social choice, and the study of complex adaptive systems to develop new ways of thinking about agency, value, and collective alignment in increasingly automated societies.
Last Update 2026/09/08Back to Member List.
