Value sensitive design in artificial intelligence: ethical issues in the case of a smart traffic signal control system

Abstract

Artificial Intelligence (AI) systems that collect huge amounts of data, raises a number of ethical concerns such citizens privacy, data security, fairness and accountability among others in societies they are deployed. Like any other technology, AI technologies are mainly imported from western economies by emerging economies, such as Africa. The question is, will an AI system built to conform to ethical requirements in one society unequivocally be accepted in another society with varying ethical requirements. This research attempts to establish whether a Value Sensitive Design (VSD) approach can address ethical issues that arise in the implementation of AI enabled IoT systems with machine learning capabilities such as Smart Traffic Signal Control (STSC) system designed for Africa (Zambia). The study identifies major stakeholders and ethical issues that arise in implementing an AI system. I conduct literature review on the philosophy of AI, ethics pertaining to AI and the major technological concepts incidental to the implementation of an AI system and how VSD can be used to implement an AI system that aligns with ethical requirements from stakeholders. In addition, I use mixed method approach to collect data from stakeholders. I compare the empirical findings with theories in literature and thereafter, make recommendations. The developing trend of major ethical issues are privacy, security and safety while most informants recommend government as a major stakeholder in addressing ethical issues in the implementation of AI systems. VSD is widely approved as a design approach for embedding values across all stages of AI systems’ lifecycle.