An interdisciplinary conceptual study of Artificial Intelligence (AI) for helping benefit-risk assessment practices: Towards a comprehensive qualification matrix of AI programs and devices (pre-print 2020)

AI needs a shared definition across disciplines: The paper identifies significant discrepancies between how psychology, engineering, ethics, and law define "artificial intelligence," highlighting that stakeholders lack a common language for discussing AI systems. A unified framework is essential for consistent assessment and regulation.

A "qualification matrix" can help assess AI risks and benefits: The authors propose a structured matrix tool that combines technical, ethical, and legal criteria to evaluate AI systems more precisely. This approach allows organizations to systematically identify which computing tools qualify as AI and what specific risks they present.

Context matters for AI assessment: Rather than applying one-size-fits-all standards, the paper advocates for a "flexible and scalable" qualification approach that adapts to different technological contexts and development stages, recognizing that AI systems vary significantly in their capabilities and potential harms.

Responsible AI development requires interdisciplinary collaboration: The research demonstrates that meaningful AI governance cannot come from a single discipline alone—psychology, engineering, ethics, and law must work together to create comprehensive assessment practices that address technical capabilities alongside societal impacts.

Benefit-risk assessment should be formalized in AI development: The paper positions its qualification matrix as a practical tool for "technological monitoring and regulatory compliance," suggesting that structured benefit-risk evaluation processes should become standard practice among AI developers and stakeholders.