Autonomy: Life and Being

This paper uses robot experience to explore key concepts of autonomy, life and being. Unfortunately, there are no widely accepted definitions of autonomy, life or being. Using a new cognitive agent architecture we argue that autonomy is a key ingredient for both life and being, and set about exploring autonomy as a concept and a capability. Some schools of thought regard autonomy as the key characteristic that distinguishes a system from an agent; agents are systems with autonomy, but rarely is a definition of autonomy provided. Living entities are autonomous systems, and autonomy is vital to life. Intelligence presupposes autonomy too; what would it mean for a system to be intelligent but not exhibit any form of genuine autonomy. Our philosophical, scientific and legal understanding of autonomy and its implications is immature and as a result progress towards designing, building, managing, exploiting and regulating autonomous systems is retarded. In response we put forward a framework for exploring autonomy as a concept and capability based on a new cognitive architecture. Using this architecture tools and benchmarks can be developed to analyze and study autonomy in its own right as a means to further our understanding of autonomous systems, life and being. This endeavor would lead to important practical benefits for autonomous systems design and help determine the legal status of autonomous systems. It is only with a new enabling understanding of autonomy that the dream of Artificial Intelligence and Artificial Life can be realized. We argue that designing systems with genuine autonomy capabilities can be achieved by focusing on agent experiences of being rather than attempting to encode human experiences as symbolic knowledge and know-how in the artificial agents we build.