Ethics of Artificial Intelligence: Experimenting with Ethics and Autonomous Vehicle AI Principles in Gameplay

Abstract

This project investigated whether ethical AI can change gameplay to be more immersive, engaging and unpredictable. The game was developed using Behaviour Trees, A* Pathfinding and Finite State Machines to model ethical AI agents. This approach introduced ethical principles from philosophy and current autonomous systems to gameplay. It gives a new experience to players where their decisions and behaviours interact with AI on a level which is fundamental to gameplay. Player interactions with AI agents have consequences, but in a way which is less predictable than standard karma systems. For example, AI agents may take on different ethical world views in different playthroughs of the game, and will also respond probabilistically, but in a non-determinate manner, to player interactions. In dynamically experiencing these changes during play the player receives these signals and may choose to change their behaviour, in the hopes of achieving a more favourable outcome. The concluding results showed that the inclusion of ethical AI increased player enjoyment and improved the dynamics of gameplay.