'DOING SCIENCE' IN A MAGIC BLACK BOX: SCIENTIFIC INTEGRITY VERSUS PRODUCTIVITY IN ACADEMIC PUBLICATIONS WITH GENERATIVE ARTIFICIAL INTELLIGENCES

Abstract

Since the launch of ChatGPT in 2022, a rapid proliferation of generative AI applications has been observed, bringing a promise of revolution in various fields, including science. These technologies are increasingly being integrated into academic studies to assist in text construction, data analysis, and knowledge extraction, enhancing scientific production and interpretation. However, as these tools advance, discussions about the use of these "Black Boxes" in scientific practice arise. The concern revolves around the limitations and ethical issues surrounding their use, raising debates such as academic integrity, transparency, reproducibility, and reliability of results. There is also concern about equity in access to and use of these tools, as well as the centralization of power in the hands of a few companies that develop these technologies. Open Science currently has guiding principles that can help shape good practices in this area. In light of this scenario, the present work seeks to explore the question: How are Generative Artificial Intelligences aligning with the guiding principles of Open Science?