Researchers at Georgia Tech have developed robots that can lie. The robots are equipped with cameras and are programmed to play hide and seek. The hider robot is able to use deception, something that is unknown to the seeking robot. The researchers claim that the need for a robot to use deception will be rare, but is potentially useful. A search and rescue robot may need to lie to a panicked victim. A robot in a war zone may be able to mislead an army or lie to the enemy if captured. The work that the GA Tech researchers have done builds on the work that Swiss researchers did in 1997 that proved that robots can learn to lie in certain situations.
Teaching robots to lie could turn out to be not a wise decision. The HAL 9000 from 2001: A Space Odyssey was programmed to lie and ended up trying to kill all of the astronauts. This, unfortunately, is what many people believe will happen if we allow robots to become too advanced. The researchers in charge of the project say that they recognize that people are leery of robots, but that deception is not necessarily wicked. We are getting closer to true AI in developing a machine that can deceive others, something that we do almost all the time, yet I do not think that robots should be able to lie. If they are programmed to communicate, they must do exactly that. The robot cannot communicate properly with people if it is lying. If I have trained a robot to perform a certain task, I want it to do that task. I do not want it to have the capacity to disobey or deceive me. I wan to be able to trust the machine that I have created and programmed.
No comments:
Post a Comment