Dependence and Sovereignty in Artificial Intelligence Technologies

an Analysis Based on the Concepts of Primitive Data Accumulation and Data Processing Inequality

Abstract

This research proposes a theoretical framework for understanding the relationship between digital sovereignty and Dependency Theory, using concepts such as primitive data accumulation and Data Processing Inequality. The study investigates how the current neoliberal model of artificial intelligence (AI) technology development is based on exploitative relations between the Global North and South, emulating colonial dynamics. This hypothesis is discussed by scholars in the digital field, through concepts such as digital colonialism. The aim is to revive a geopolitical reading of Dependency Theory, particularly in the contemporary interpretations by Theotonio Dos Santos (2020) and Claudia Wasserman (2022). The research suggests that Dependency Theory aligns with the concept of primitive data accumulation, as outlined by Lippold and Faustino (2022), which draws an analogy between the current data economy and the initial accumulation of productive assets in individualized private property, as identified by Marx. To construct this dialogue, the study integrates Dependency Theory with the principle of Data Processing Inequality, which states that any data transformation process cannot increase the information about the measured variable (Beaudry, Renner, 2012). This principle explains how the production of the "new" in AI technologies necessarily depends on the continuous input of new data, and proposes an explanation for the colonial-like expansion of AI technologies in the Global South as a consequence of the need for new data to support the primitive accumulation of capital.