It’s Only a Bot! How Adversarial Chatbots can be a Vehicle to Teach Responsible AI

We are currently witnessing an ever-growing entanglement of intelligent technology with people in their everyday lives, creating intersections with ethics, trust, and responsibility. Understanding, implementing, and designing human interactions with these technologies is central to many advanced uses of intelligent and distributed systems and is related to contested concepts, such as various forms of agency, shared decision-making, and situational awareness. Numerous guidelines have been proposed to outline points of concern when building ethically acceptable artificial intelligence (AI) systems. However, these guidelines are usually presented as general policies, and how we can teach computer science students the needed critical and reflective thinking on the social implications of future intelligent technologies is not obvious. This chapter presents how we used adversarial chatbots to expose computer science students to the importance of ethics and responsible design of AI technologies. We focus on the pedagogical goals, strategy, and course layout and reflect how this can serve as a blueprint for other educators in broader responsible innovation contexts, e.g., nonchat AI technologies, robotics, and other human-computer interaction (HCI) themes.