Principles for the future development of artificial agents

摘要

A survey of popular, technical and scholarly literature suggests that autonomous artificial agents will populate the future. Although some visions may seem fanciful, autonomous artificial agents are being designed, built, and deployed in a wide range of sectors. The specter of future artificial agents - with more learning capacity and more autonomy - raises important questions about responsibility. Can anyone (any humans) be responsible for the behavior of entities that learn as they go and operate autonomously? This paper takes as its starting place that humans are and always should be held responsible for the behavior of machines, even machines that learn and operate autonomously. In order to prevent evolution to a future in which no humans are thought to be responsible for the behavior of artificial agents, four principles are proposed, principles that should be kept in mind as artificial agents are developed.