RESEARCH AND ARTIFICIAL INTELLIGENCE. ETHICAL AND LEGAL CHALLENGES

AI-Research binomial relationship raises some ethical challenges, which should not be ignored by researchers. These challenges are not limited to issues arising from the use of AI in scientific research carried out in the designated “scientific research system” or in the various sectors of applied research. The conditions under which AI research is carried out, an industry dominated by a limited number of large technological companies, and which is research intensive, must also be the subject of attention. In fact, the industry has taken the lead in AI research and both dimensions are increasingly intersected, and that is further enhanced by general purpose AI models. The flip side of this reality is a lower diversity of research promoted by industry in relation to research promoted by academia, including with respect to gender. Besides the most obvious challenges, such as the risk of biases, the respect for privacy or the accountability of “black  box” systems, there are challenges in the development of AI research that are less obvious, namely an ambivalence regarding sustainability, labor exploitation in the training of AI systems or the disrespect for human creation. The use of AI in scientific research has obvious potential for acceleration and accuracy, and it is already significant. Both the EU, the US and China recognize this and have advanced initiatives to ensure this potential is realized. However, some old problems, namely guaranteeing research integrity, critical thinking, and even "apocalyptic" risks, such as of replacement of human researchers, have been highlighted. The answer to the challenges, risks and dilemmas lies in permanent scrutiny and a multi-stakeholder, multi-level regulatory approach throughout the entire cycle of AI research and use, according to human values.