Grasping the Shape of Ethical Dilemmas: Towards an Educational Dilemma Generator

The incorporation of data science and artificial intelligence (AI) technologies in different fields and areas provide many benefits in automation, resource consumption, and efficiency. However, the deployment of information systems that can exhibit autonomous behavior—and, thus, that can make decisions that have an effect on their users—may need to be prepared to face some ethical dilemmas. Those dilemmas cannot often be completely covered by following an ethics by design approach, which is based on foreseeing, beforehand, all potentially detrimental scenarios. In order to allow autonomous systems to act ethically and fairly, artificial morality techniques are a promising way to go. Our goal is to use artificial morality systems to support the incorporation of AI in eLearning environments. However, before implementing systems that are able to identify and react to ethically relevant scenarios, there is the need to understand what makes an educational situation be considered ethically relevant and why. In order to gain this knowledge, this work presents a preliminary prototype that automatically generates hypothetical educational situations that may involve (or not) potential ethically relevant decisions: the Educational Dilemma Generator. The work characterizes ethical dilemmas in education, presents the design of the Educational Dilemma Generator, and shows the results of the preliminary tests conducted. The contribution of the research is to create tools to advance in the knowledge about what ethical dilemmas in education are and what characterizes them.