Published May 18, 2026 | Version v2

Operationalizing AI Ethics in the Brazilian Public Sector: An Empirical Study of Practitioners' Motivations, Perceptions, and Expectations

Authors/Creators

Description

Background: The adoption of Artificial Intelligence (AI) systems in the Brazilian public sector has accelerated, yet a persistent gap remains between high-level ethical principles defined in regulatory frameworks and their effective translation into software development and organizational practices.  Aims:  This study empirically investigates the socio-organizational factors that influence AI ethics operationalization in the Brazilian public sector and examines how these factors are reflected in practitioners' motivations, perceptions, and expectations.  Method:  We conducted a diagnostic survey with civil servants and public employees involved in the development, deployment, or governance of AI systems. The instrument comprised 15 five-point Likert items distributed across three analytical dimensions (Motivation, Perception, and Expectation) and three open-ended questions, yielding 123 valid responses. Quantitative analysis included descriptive statistics, Cronbach's alpha, Spearman's rank correlation, and subgroup analyses, while qualitative data were analyzed through thematic categorization of 181 open-ended responses. Results:  Extrinsic motivators (legal compliance, transparency, and reputational risk) exceeded intrinsic organizational drivers by approximately 1.2 points on the Likert scale. Furthermore, 75.6% of respondents reported difficulties translating ethical principles into technical requirements or code, 61.8% perceived a discourse-practice gap, and only 22.0% perceived alignment among technical, legal, and managerial stakeholders. Strong consensus (87.8--93.5%) emerged around five operationalization mechanisms: practical frameworks, auditing tools, interdisciplinary committees, continuous training, and shared accountability. Conclusions: The findings confirm a persistent discourse-practice gap and suggest that AI ethics operationalization requires an integrated ecosystem of technical artifacts, governance mechanisms, and capacity-building rather than isolated interventions, informing 18 preliminary requirements for a context-specific ethical framework.

Files

survey_questions.pdf

Files (340.4 kB)

Name Size
md5:e98b9b771c8ec9d1b875fd2c81457c95
28.3 kB Download
md5:58f1780f10b79fc97674602039a72a7c
235.4 kB Preview Download
md5:942f12a266861c7708e6127df20efc52
76.8 kB Preview Download