The problem of artificial phronesis is one that has to be solved by the designers of social robots—otherwise their project to bring robots into our daily lives will fail. At its core, phronesis is a term used by Aristotle to describe the skill some people have for using practical reasoning to navigate the ethics of social interactions with proficiency and excellence. Phronesis is an ability that most humans have but it is also one that requires practice to master. Once attained this skill is used by an ethical agent to assess the appropriately virtuous actions one might take in a given social situation. Since there are a vast number of actions any agent might do in any given situation, this skill is used to quickly separate the many inappropriate actions that might be done from the few just and right actions that are possible in the situation. The problem is that the right actions are often more difficult to accomplish and costly to the agent whereas the wrong actions can be easy and profitable to the agent. For Aristotle, this skill is not a matter of logical deduction, such as, “when in situation x, then always do action y.” Instead, it is a skillful practice that requires the discernment of the nuances involved in any real-life situation. Since any real-life situation of consequence is one that is encountered only once in in a person’s lifetime, there is no information from past actions or situations that will be wholly sufficient for deducing a proper reaction.
Social robots have only their programing and machine learning to rely on for deducing their reactions to any given social situation. If phronesis is a real ability humans have and it is not something that can be programed or learned simply form past experience, then the social robot’s lack of that capacity will be an insurmountable barrier to the machine’s ability to enter into meaningful relations with human agents.
We will explore this problem and look at a possible solution inspired by the philosopher John Dewey where he expands on Aristotle’s conception of phronesis and tries to ground this capacity in nature. If he is successful, then phronesis would not be the exclusive capacity of human agents, but other kinds of agents might be capable of it as well. Might that expanded set of agents include social robots? Let’s find out.
