A framework for ethical artificial intelligence - from social theories to cybernetics-based implementation

Abstract

The proposed work aims to develop an ethical framework for the implementation of artificial intelligence (AI). The present work changes the discussion from 'What is AI ethics' to 'how AI system developers can implement AI ethics'. The current work deploys cybernetics principles to address the challenges pertaining to AI ethics implementation. With the help of two pillar elements of cybernetics, i.e., man and machine, AI ethics principles have been elucidated in the present work. The study demonstrates that cybernetics provides a different dimension to implement AI ethics principles and dispenses a basis to deploy already existing AI ethics principles. The combination of cybernetics theory and AI ethics principles serves as a firm foundation for implementing AI ethics. The present work provides a comparative study of IBM principles for AI ethics, Japanese Society of Artificial Intelligence's (JSAI's) AI ethics principles, and the proposed cybernetics-based AI ethics framework to provide holistic visualisation.