Agência e responsabilidade no contexto de sistemas autônomos artificiais
Carregando...
Arquivos
Data
2020-12-18
Autores
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
Esta pesquisa tem por objetivo investigar, no contexto das tecnologias informacionais emergentes, a possibilidade de sistemas autônomos artificiais serem considerados agentes e, em particular, agentes morais aos quais cabe atribuir algum grau de responsabilidade. A pergunta que guiará esta investigação pode ser assim formulada: Seria legítimo atribuir algum grau de responsabilidade a sistemas autônomos artificiais, levando em consideração que estes sistemas são produzidos e programados por sistemas autônomos naturais e intencionais? Para tanto, explicitaremos se, e em que condições, podemos classificar um agente artificial como autônomo e o que entendemos como conduta responsável. Procuraremos mostrar que sistemas artificiais que forem capazes de extrapolar sua programação inicial, aprender com a própria experiência e alterar o rumo de sua ação como resultado de tal aprendizagem podem vir a ser considerados elementos constituintes de uma rede de agentes intencionais. Tal atribuição estaria baseada na tese de que agentes artificiais, instanciadores de intencionalidade derivada, teriam também corresponsabilidade derivada sobre as possíveis consequências de suas performances. Por outro lado, recusar a possibilidade de agentes artificiais poderem ser considerados corresponsáveis pelas consequências de suas performances implicaria não considerar devidamente a capacidade de tais sistemas alterar o rumo de sua conduta como resultado de sua própria experiência. Assim, entendemos que a clássica relação entre agência e responsabilidade moral sobre as consequências da conduta precisa ser revista de modo a incluir a hipótese de agência artificial e suas possíveis relações com a responsabilização moral no contexto das tecnologias contemporâneas.
This research aims to investigate, in the context of emerging information technologies, the possibility of artificial autonomous systems considered to be agents and, in particular, moral agents to which it is appropriate to attribute some degree of responsibility. The question that will guide this investigation can be formulated as follows: Would it be legitimate to attribute some degree of responsibility to artificial autonomous systems, taking into account that these systems are sought and programmed by natural and intentional autonomous systems? For that, we will explain if, and under what conditions, we can classify an artificial agent as autonomous and what we understand as responsible conduct. We will try to show that artificial systems that enable them to extrapolate their initial programming, learn from their own experience and change the course of their action as a result of such learning can come to be considered as constituent elements of a network of intentional agents. This must be based on the possibility that the artificial agents, instantiators of derived intentionality, also derived co-responsibility for the possible consequences of their performances. On the other hand, to refuse the possibility that artificial agents can be considered co-responsible for the consequences of their performances, would imply that we do not properly consider the capacity of such systems to change the course of their conduct as a result of their own experience. Thus, we understand that the classic relationship between agency and moral responsibility on the consequences of conduct needs to be revised to include the hypothesis of artificial agency and its possible relationships with moral accountability in the context of contemporary technologies.
This research aims to investigate, in the context of emerging information technologies, the possibility of artificial autonomous systems considered to be agents and, in particular, moral agents to which it is appropriate to attribute some degree of responsibility. The question that will guide this investigation can be formulated as follows: Would it be legitimate to attribute some degree of responsibility to artificial autonomous systems, taking into account that these systems are sought and programmed by natural and intentional autonomous systems? For that, we will explain if, and under what conditions, we can classify an artificial agent as autonomous and what we understand as responsible conduct. We will try to show that artificial systems that enable them to extrapolate their initial programming, learn from their own experience and change the course of their action as a result of such learning can come to be considered as constituent elements of a network of intentional agents. This must be based on the possibility that the artificial agents, instantiators of derived intentionality, also derived co-responsibility for the possible consequences of their performances. On the other hand, to refuse the possibility that artificial agents can be considered co-responsible for the consequences of their performances, would imply that we do not properly consider the capacity of such systems to change the course of their conduct as a result of their own experience. Thus, we understand that the classic relationship between agency and moral responsibility on the consequences of conduct needs to be revised to include the hypothesis of artificial agency and its possible relationships with moral accountability in the context of contemporary technologies.
Descrição
Palavras-chave
Filosofia da tecnologia, Tecnologias informacionais, Sistemas artificiais, Responsabilidade, Ética da informação, Philosophy of technology, Information technologies, Artificial systems, Responsibility, Information ethics