Zac Cogley (Northern Michigan University)
Thursday, April 21th, 5:00-6:30 ASRC 230
Future Autonomous Weapons Will Make Moral Judgments
My aim in this paper is to provide a broad sketch of how the moral judgment of autonomous weapons systems (AWS) might work by appealing to recent results in autonomous computing. I show that recent results in deep computing - specifically, the success of the program AlphaGo at mastering the game Go - suggests that future AWS will be able to approximate the human capacity for moral judgment and allow AWS to make targeting decisions for the right reasons. AlphaGo couples a search tree - a set of rules that describe all possible moves in the game - with two deep neural networks. I argue that one network allows AlphaGo to make judgments about the strength of different positions. The other allows AlphaGo to see the reasonable moves given the current state of the game. While we are a long way from the creation of an AWS that will actually ghave the relevant moral capabilities, we have no principled reason to think that one cannot be developed by extrapolation from the features of AlphaGo and other deep learning systems.
Other Upcoming Talks:
May 12th at 5pm: Patrick Smith (National University Singapore)
June 2-3 TBD: Neil Sinhababu (National University Singapore)
The Socratic Society
The Socratic Society of Portland State University is a venue for Philosophy faculty and philosophers from outside the University to present papers or speak on various topics of interest. Meetings for the 2013-14 academic year will be announced as they are scheduled.
Please check periodically for upcoming events, and if you are interested in presenting or participating with the Socratic Society at PSU, please contact Avram Hiller at email@example.com.