Thursday, November 1, 2018 4:10pm
About this Event
6A E Packer Ave, Bethlehem, PA 18015
https://cogsci.cas2.lehigh.edu/Cognitive Science Program, Computer Science and Engineering Department, Center for Ethics
How to/Not to Engineer Morally Correct Machines
Selmer Bringsjord
Rensselaer Polytechnic Institute
Ethics as a systematic field has been in operation for at least two millennia; it centers around such questions as: What ought we to do? What are we prohibited from doing? What is heroic? Why ought we to do what we ought to do? And so on. Notice the occurrences of ‘we’ in these questions. As humanity approached and moved into the third millennium, something radically new arrived on the scene: namely, machine ethics. Here, the idea
isn’t our moral status and condition; rather, the focus is on the design and engineering of machines (including robots, autonomous weapon systems, etc.) that themselves do what ought to be done, don’t do what is forbidden, and — when relevant circumstances arise — behave heroically. I explain in this talk (1) why engineering morally correct machines is something we must do, (2) that such engineering, if based on the much-hyped “machine learning” of today will fail and likely get most of us killed, but (3) that there is a way to engineer morally correct machines, and thereby prevent catastrophe. This way is explained.