A Multi-Aspectual Requirements Analysis for Artificial Intelligence for Well-being

Technology, and specifically Artificial Intelligence, is transforming society in profound bivalent ways with positive and negative impacts. There is a recognition that AI offers immense potential for improving human well-being, which has been formulated at length from many perspectives, including the economic, social, psychological, developmental, and health perspectives. Critical questions with regards to these AI for well-being systems are, “What does success as far as employing AI to support human well-being look like?” and perhaps more importantly, “How do we inform system development processes to ensure the said success?” AI ethical frameworks and design guidelines have been formulated that address these questions by articulating some notion of successful or desired AI. Further, approaches such as participatory design, value-sensitive design, and goal-oriented requirements engineering aim to connect system development processes with high-level contextual concerns and non-functional softgoals. However, to establish the effectiveness of these instruments for informing AI for well-being solutions, it is necessary to have normative analysis and evaluation frameworks that recognize the innate complexity and multidimensionality of human well-being. This paper presents Herman Dooyeweerd's philosophy as one such framework and provides a formulation of well-being informed by his Theory of Aspects and his “simultaneous realization of norms” principle. The paper then presents a multi-aspectual analysis of AI Ethics frameworks to illustrate and suggest a mechanism that can be employed across diverse contexts and domains for requirement analysis and requirements refinement for technology solutions in general, but specifically for AI for well-being solutions.