BETWEEN SOVEREIGNTY AND EFFICIENCY: THE BRAZILIAN ARTIFICIAL INTELLIGENCE PLAN AS A TECHNICAL CODE

o Plano Brasileiro de Inteligência Artificial como código técnico

Abstract

This article analyzes the Brazilian Artificial Intelligence Plan (PBIA) as the institutionalized expression of an emerging technical code, in light of Andrew Feenberg's Critical Theory of Technology. The objective is to investigate to what extent the plan institutionalizes a code that guides AI governance in Brazil, identifying its technical assumptions and social values. To this end, a qualitative approach of critical documentary analysis is adopted, grounded in the philosophy of technology and critical qualitative research, which understands social phenomena as historical and normative constructions permeated by power relations. The PBIA is treated as a normative sociotechnical artifact, analyzed through a three-dimensional lens that articulates the technical code (Feenberg), the politics of artifacts (Winner), and moral mediation (Verbeek). The results show that, although the PBIA mobilizes an ethical-inclusive discourse centered on the pillar "AI for the Good of All," its structure reveals the predominance of an instrumental rationality. This logic subordinates social principles to criteria of efficiency and competitiveness, where the "good of all" is mediated by functional performance. Budgetary asymmetry corroborates this finding: 59.9% of investments are allocated to business innovation, compared to 0.5% for democratic regulation and 5% for citizen education. Values such as diversity and inclusion tend to be absorbed by a logic of technical compliance, resulting in weak moral mediations incapable of reconstituting the dominant logic. Despite the tendency towards depoliticization, the investigation identifies normative gaps and potentials for subversive rationalization within the plan itself, such as commitments to cultural diversity and environmental sustainability. It is concluded that the PBIA stabilizes hegemonic interests under the appearance of operational neutrality but remains a field of political dispute. The study emphasizes that AI governance in Brazil requires overcoming the dichotomy between technique and politics. Its effective democratization will therefore depend on creating deliberative arenas capable of redistributing technical power and embedding citizen participation at the heart of defining the country's technological directions.