GENERATIVE ARTIFICIAL INTELLIGENCE IN SOCIETY: POTENTIALS, RISKS AND RESPONSIBLE GOVERNANCE
TL;DR
This article examines GenAI applications, limits, and governance requirements, with particular attention to Brazil, and proposes the GERA-BR Framework, structured around seven connected movements: govern, frame, safeguard data, assess, conduct human review, trace, and learn.
Abstract
In a short period, generative artificial intelligence (GenAI) moved beyond specialist settings and entered everyday practices of study, communication, management, and knowledge production. The shift created opportunities for productivity and personalization, but it also brought difficult questions concerning reliability, data protection, copyright, and information integrity. This article examines GenAI applications, limits, and governance requirements, with particular attention to Brazil. The study combines a non-exhaustive integrative review with documentary analysis. Scientific databases and institutional sources were consulted between May and July 2026; after deduplication and screening, 35 documents formed the corpus. A clear pattern emerged. Benefits were more consistent when tasks were well defined, acceptance criteria were explicit, and human oversight was meaningful. Their magnitude varied with data quality, task design, and user experience. Risks involving hallucinations, bias, anthropomorphism, misinformation, opacity, cognitive dependence, and environmental costs remain. As an applied contribution, the article proposes the GERA-BR Framework, structured around seven connected movements: govern, frame, safeguard data, assess, conduct human review, trace, and learn. Responsible adoption, the study concludes, requires evidence, transparency, rights protection, capacity building, and continuous monitoring. As complementary implementation instruments, the study presents maturity levels, an institutional roadmap, and indicators for monitoring quality, risk, and learning.