Artificial intelligence in the long view: from mechanical intelligence to cyber-social systems

Abstract

AbstractArtificial intelligence (AI) is emerging as a defining technology of our time, a source of fear as often as inspiration. Immersed in its practicalities, rarely do we get to ask the question, what is it? How does it impact our lives? How does it extend our human capacities? What are its risks? What are its limits? This paper is a theoretical and historical overview of the nature of binary computing that underpins AI and its relations with human intelligence. It also considers some philosophical questions about the semiotic or sense-creating work of computers. Our argument proceeds in five steps. We begin with an historical background: since Ada Lovelace, we have wondered about the intelligence of machines capable of computation, and the ways in which machine intelligence can extend human intelligence. Second, we ask, in what ways does binary computing extend human intelligence and delimit the scope of AI? Third, we propose a grammar with which to parse the practical meanings that are enabled with and through binary computing. Through this discussion, we raise the question of ontology as a counter-balance to what we will argue has been an over-emphasis on the instrumental reasoning processes of the algorithm. Fourth, we situate binary computing in the context of broad developments in modern societies which we characterize as a series of systems transitions: from industrial, to informational, to a new phase that we term “cyber-social.” Finally, we explore the risks inherent in a pervasively cyber-social system. These are narrowly captured in the technical domain, “cybersecurity.” We set out to reconceive this problem framework as the location for a potential solution, supplementing analyses of cybersecurity risk with a program of cyber-social trust.