Focuses of Recent AI Ethics Guidelines and Personal AI Agent

Abstract

The notion of singularity has invoked us the idea that AI will have the same or more intellectual abilities as human beings have, and finally become a threat to us. This has triggered the momentum for research of AI ethics around 2016, and the communities of AI researchers started to create AI ethics guidelines which have been publicly available now. In this paper, we focus on the important and influential AI ethics guidelines published domestically and overseas from 2017 to 2019 and analyze how each AI ethics guideline has dealt with ethical themes such as AI control, human rights, fairness, non-discrimination, transparency, accountability, trust, abuse, misuse, privacy, AI agents, safety, SDGs, education, antimonopoly and coordination, policy, military use, legal status, and well-being. The time sequence of the publication of the various AI ethics guidelines is used here to explore changes in the content of AI ethics, and at the same time, clarify points to be noted when developing AI technology and AI application systems. Considering the expected readers of each of these guidelines, it reveals clearly what are the intention of the organizations that created these AI ethics guidelines.

Next, we describe the conceptual design of a proxy software for collecting, managing, and protecting personal data, that we call “personal AI agent”. The basic concept was proposed in IEEE EAD version 2 and 1^st edition among the AI ethics guidelines. The personal AI agent consists of two databases of the data subject's personal data and the terms and conditions about the usage of them. We also indicate the issues to be considered when applying the personal AI agent to the management of digital heritage left after the death of the data subject.