Ethical Considerations in Artificial Intelligence Courses

22 AI MAGAZINE Artificial intelligence (AI) is one of the most ambitious scientific and engineering adventures of all time. The ultimate goal is to understand the mind from a new perspective, and to create AIs1 capable of learning and applying intelligence to a wide variety of tasks: some as robots able to take action in our physical and social world, and some as software agents that make decisions in fractions of a second, controlling huge swaths of the economy and our daily lives. However, the power and reach of these AIs makes it necessary that we consider the risks as well as the rewards. In thinking through the future of AI, it is useful to consider fiction, especially science fiction. From Frankenstein’s monster and Hoffmann’s automata to Skynet and Ex Machina, fiction writers have raised concerns about destruction that could perhaps be unleashed on humanity by the autonomy we confer on our technological creations. What is the underlying concern that inspires so many variations of this story? Do these storytellers (and their audiences) fear that AIs, by definition, cannot be trusted to act in a way that does not harm the society that creates them? Or do they fear that the people in charge of designing them are making the wrong