Towards a Human Rights-Based Approach to AI: Case Study of Apprise

Abstract

Frontier technologies such as Big Data and Artificial Intelligence (AI) are hailed to improve decision-making by reducing and even mitigating human biases. The emergence and rapid adoption of these technologies, particularly in optimization of services and provision of key analytics and insights was justified by the widespread benefits of AI to democratize intelligent software for all. Yet, recent studies have brought to light cases where AI has perpetuated existing biases and deepened inequalities, contributing to the further marginalization of specific groups in society. Despite the opportunities that AI offers, it also poses new threats to human freedom, fairness, non-discrimination, privacy, and security; leaving questions regarding the human rights implications of AI unaddressed. This paper proposes the use of international legal frameworks such as the International Bill of International Human Rights (including the Universal Declaration of Human Rights) to assess the human rights impacts of AI system. To ground the discussion, we present a case study to assess the human rights implications of Apprise, a multi-lingual expert system for screening potential victims of human trafficking and forced labor, piloted in Thailand. Drawing on amplification theory, we highlight that AI systems are not deployed in neutral systems, and that pre-existing inequalities and “unfreedoms” can be aggravated if not addressed. We argue for a balanced view of the potential of AI systems, cognizant of both the positive and negative intentions of users of such technologies.