Value Creation and Value Capture in AI: A Triple Helix Model

Abstract

The idea of ‘value chain’ faces limitations as an effective concept to address the complex mechanisms of value creation and value capture mechanisms in Artificial Intelligence. Starting with a review of Michael Porter’s original concept, this paper explores the use of the term in the AI context, highlighting the need for a more robust theoretical foundation to prevent reductionist accounts of value. To address the theoretical gap and make the complexity of value creation and value capture in AI tractable, we introduce the triple helix model of AI—a comprehensive analytical framework that decomposes AI into its three fundamental strands: data, architecture, and computational power. With this, AI can be understood at the intersection of these three strands in each social context, capturing different value dynamics. Our methodology advances a three-phase approach to thinking about the triple helix: (1) an ontology or list that identifies key elements and their relationships; (2) a kinematic account which portrays—in a historical way—the evolution of these components; and (3) a dynamic account, which al-lows the identification of the driving forces of such evolution. This allows the drawing-up of a comprehensive and faithful account of AI stagnations and booms while opening the door to reading AI evolution as a story of power.