A thematic synthesis of ethics principles in artificial intelligence

Abstract

In an era marked by rapid advancements in artificial intelligence (AI), the ethical dimensions of AI development and deployment have become increasingly pivotal. As AI technologies permeate diverse sectors, the need for a comprehensive understanding of the ethical principles governing their use has intensified. This research employs reflective thematic analysis to scrutinise the ethical landscape of AI, to discern consensus among stakeholders and evaluate the practicality of implementing ethical principles. Leveraging the critical-systems-heuristics framework, the study explores implicit assumptions, power dynamics, and contextual intricacies for a nuanced analysis. Data from 156 entities form the basis for a qualitative thematic synthesis, revealing motivations, control mechanisms, knowledge sources, and legitimacy factors guiding AI-ethical principles. Key findings spotlight the prevalence of ethics documents in the private sector, driven by market competition, corporate social responsibility, regulatory compliance, and stakeholder expectations. Europe and North America have emerged as leaders in document publication, reflecting their technological prowess. Government agencies uniquely emphasise transparency. Variations in prioritised principles across stakeholders unveil distinct motivations aligned with organisational goals. Challenges impeding AI-ethics implementation encompass vague principles, global regulatory disparities, data-privacy concerns, and resource limitations. The study unravels worldviews which shape AI ethics, with private organisations valuing human-centricity, accountability, and legitimacy through representation and consensus. The outcomes contribute theoretical insights and practical recommendations, guiding the responsible development of AI technologies.