PRIVACIDADE POR DESIGN NO DESENVOLVIMENTO DE INTELIGÊNCIAS ARTIFICIAIS: TRANSPARÊNCIA E PRIVACIDADE DIANTE DO DESAFIO DA CAIXA PRETA

Abstract

The significant expansion of Artificial Intelligence (AI) and the massive volume of data generated daily present certain challenges. Prominent among these is the "black box" problem, resulting from algorithmic opacity and a lack of understanding regarding the logic of intelligent systems. This research aims to analyze how Privacy by Design (PbD) concepts can guide developers of intelligent algorithms (AI) to ensure transparency and data privacy from the project's inception. This study conducted qualitative exploratory-descriptive research, analyzing 25 relevant documents from scientific literature to identify connections between Privacy by Design and transparency in AI. The results indicate that the seven principles of PbD can guide the development of these algorithms so that transparency becomes a positive, intrinsic functionality of the system. Furthermore, the use of tools such as Explainable AI (XAI) and LIME can transform opaque systems into explainable algorithms for both users and developers. Ultimately, this allows for the reduction of vulnerabilities and ethical risks, establishing trust in the digital age.