Gunsberg on AI and Responsibility

Patrick Günsberg (Uppsala University - Faculty of Law) has posted Some Thoughts on Artificial Intelligence (AI) Systems and Responsibility on SSRN. Here is the abstract:

 

With the growing autonomy of AI systems, one is left asking who should be held accountable when such systems go wrong. Autonomous weapons could, for example, engage in activity that makes one think of war crimes while pricing algorithms could be detrimental to consumers. When considering the responsibility of individuals and firms for such activity one would do well to bear in mind that humans create algorithms and determine how algorithms are used. It appears that one should not rule out the possibility of holding programmers accountable for the actions that AI systems take. Nevertheless, there seems to be definite appeal to the suggestion that users should be responsible for the actions of AI systems. Yet the arguable difficulty of demonstrating the causal chain of events may speak in favor of introducing strict liability – compensating injured individuals seems important even if a causal link cannot be established.