Abstract

Autonomous and intelligent systems and services that use narrow artificial intelligence technologies such as statistical learning and limited inferencing (AISSN) are pervasive in our lives and industry. These systems, very far from having human-like intelligence, will offer significant potential for doing social good, achieving productivity gains and advancing science and engineering. However, AISSN systems can have unanticipated and harmful impacts. This chapter highlights the ethical challenges of AISSNs using three diverse and pervasive examples: Internet of Things, conversational AI, and semi-autonomous vehicles. We contend that AISSNs will be the norm for the foreseeable future and that artificial general intelligence will not develop anytime soon. The ethical challenges of AISSNs are addressable using human-centred “Ethical Design”, the use of widely accepted moral standards of right and wrong to guide the conduct of people in the ideation, design, development, and deployment of AISSN systems. Depending on the problem domain, multidisciplinary teams of computer scientists and engineers, sociologists, economists, ethicists, linguists, and cultural anthropologists will be required to implement humanistic design processes.