Dynamic value alignment through preference aggregation of multiple objectives

Marcin Korecki, Damian Dailisan, Cesare Carissimo

The development of ethical AI systems is currently geared toward setting objective functions that align with human objectives. However, finding such functions remains a research ch…