Is effective regulation of AI possible?
John Danaher highlights eight potential regulatory problems.
Considers Matthew Scherer’s article ‘Regulating Artificial Intelligence Systems’.
Firstly we must be able to adequately define what we are regulating.
Local Control Problem - when an AI cannot be controlled by those legally responsible.
Global Control Problem - when an AI cannot be controlled by any humans.
During development - Discreetness, Diffuseness, Discreteness and Opacity problems.