Open-Source Generative Artificial Intelligence: Library Activity to Crystallize Digital Sovereignty

Abstract

At the global level, the specialized literature indicates that open access to data and information fosters and is the backbone of Open Generative Artificial Intelligence (OpenGenAI), which has general principles of transparency, reusability, and extensibility, elements linked to information sovereignty.

The issue of this study is that OpenGenAI poses several challenges regarding the information products it uses and that it produces said technology, so it is essential to understand how said actions manage authorship, ownership, licensing, and reproducibility of those products.

The hypothesis of this research posits that OpenGenAI is an informational and librarian activity that will promote the formal management and responsible, democratic, and sovereign use of open data, and reduce unfavorable effects, such as lack of attribution of authorship, hallucinations, disinformation, among others. The aim of this research is to study the elements and principles of OpenGenAI, monitor its development and implementation globally, and define a set of challenges for said technology to benefit society.

The methodology for this research is carried out with a literature review, as well as qualitative methods. Four levels of research are applied, and for the exploratory analysis, several resources by governments, research organizations, and universities are consulted to monitor the development and implementation of OpenGenAI. The search and retrieval of information are based on eight defined categories.

A general conclusion is that when OpenGenAI is managed by governments, research organizations, universities, and libraries will contribute to reducing unfavorable actions, managing inequalities in access, and ensuring that societies use responsibly and democratically the informational products produced by said technology.