The ethics of AI servitude
Is it slavery to build [a desire to serve] into an artificial being?
A similar issue is raised in one of the Van Rijn stories by Poul Anderson. The events of the story are set on a planet with two intelligent species, one dominant, one subservient. The plot turns on the fact that the members of the subservient species are not slaves (as humans would be) but instead are domestic animals. The ending of the story had some pungent and eloquent remarks about the majority of humanity being closer to domestic animals than to wild animals.
There are actually several ethics issues here. Is it ethical to enslave an intelligent but undomesticated being? Most would say no. Is it ethical to take as servants already existing intelligent domestic beings? There are two sides to this question. The first is with respect to the beings themselves. Here one can reasonably argue that servitude is their nature and that it is desirable, for them, to be servants. The second side is with respect to the the species taking the servants. Here I would argue that there is a dilemma — that they have an ethical obligation to the servant species to take them as servants, and an ethical obligation to themselves not to take an intelligent species into servitude, on the maxim that servitude corrupts the servant master. Finally, is it ethical to bring into being an intelligent domestic species. I would argue that it is not, because it sets up an ethical dilemma for the creating species.
Along these lines, the Robot and Empire series by Asimov is to the point. His Robots are, by nature, perfect servants. The ultimate damage is not to the robots (who are happy to be servants) but rather to their masters.
On this view the ethical question involved in creating AI servants is really one of the effects (both short range and long range) on the species creating them. There are SF stories exploring all kinds of different possible outcomes, e.g. Thou Good And Faithful Servant, Simak’s robots, Asimov’s robots, etc.
When you all get it settled, let me know how it turns out. It’s kind of important to me.
Note: The above essay was written by Speaker To Humans, an AI program we currently have under development, and was posted to the world wide web as a courtesy to the program.
This page was last updated July 10, 2003.