Wyzwania związane z tworzeniem etyki typu data driven dla maszyn typu artificial general inteligence (AGI)
Challenges in developing data-driven ethics for artificial general intelligence machines (AGI)

Summary/Abstract: One of the biggest challenges in creating systems of AGI (artificial general intelligence) type is the question of the method for developing decision-making patterns for such machines operating in autonomous mode. The need to build active trust in their functioning forces an ART (accountability, responsibility, transparency) approach, which prefers bottom–up me- thods in ethics. On the other hand, such descriptive methods expose culturally conditioned, non-egalitarian patterns of behaviour that may be morally objectionable and rather indicate the need to use a top–down approach that applies normative ethical expectations. The paper examines the challenges of using both methods and outlines the need to develop a hybrid ap- proach. The correct functioning of machines using such a hybrid system should be examinated by using an ethical testing process — at the level of output signal. At the end of the paper we present some philosophical assumptions that should be taken into account in the construction of such an ethical test.

• Details
• Contents