Ética em sistemas multiagente autônomos

Abstract

In the current moment of humanity, working to have ethics in Autonomous Multi-Agent Systems has become mandatory given the evolution of this technology in recent years. Humans have been deceiving themselves since the species began, leading us to ask if this is already replicated by computer systems, if it is possible to detect the machine's lie, if they commit deception and how to identify this attitude. This research focuses on the creation of the Severus agent with the aim of using Artificial Intelligence (AI) to supervise AI. An agent that makes in its programming a great cooperation and collaboration of the ethics considered in exact sciences with all the questions of the humanities-social sciences and that has in the implantation and implementation an evident bottom-up approach. A work designed to adjust the level of autonomy of agents in the system, essentially observing respect for human rights, the public good as paramount and the use of shared intelligence in the service of improving good coexistence. To achieve this goal, the agents work under human command, especially having the Requirements Engineer responsible for the ethical control of the systems in his great function of negotiating the possible balance with the stakeholders. Speaking in simple language, towards those who work, live and love in the most diverse societies, the primordial of ethics in this new post-human era is to consider that AI will be a fundamental pillar to define if humanity will have a future. This research concludes that to succeed and to have good ethics in Multi-Agent Systems, we must have an active participation in all decisions, learn and respect the psychology of the new diversity in teams of humans and machines and contribute to the formation of skilled leaders who lead an AI indeed cooperative and working in the service of humanity. This conclusion will always be dynamic, as the agents present great differences in behavior in the different environments, being necessary an effective work of the Requirements Engineer to adapt the matrix of results of the teams for a good option in the dominant strategy avoiding the threats that this new era of technological devices represents.