How well do artificial intelligences (SAIs) reason morally? This article offers a test of their ethical decisions and reasoning offered using the Kohlberg model of moral development. "By bridging the connection between human psychology and AI," the authors write, "we are paving a way to understand how these models work." As a computing exercise the test is informative, but the psychology of morality is much more complex than described here. And it's worth noting that a criticism of the Kohlberg model is precisely what stimulated feminist philosopher Carol Gilligan to develop an alternative approach to ethics based on care and relationships. So there's work to be done here as well. Via Mike Young, who summarizes the paper.
Today: 7 Total: 98 [Direct link] [Share]
