Altruism in Game AI: using Deep Reinforcement Learning to create AI agents which exhibit aluristic or selfish behaviours

Altruism and Selfishness: Believable AI game agents

This research aims to investigate the use of RL to create believable game agents by training them to exhibit human-like altruistic behaviour. It further aims to establish an understanding of the effect of selfish or altruistic behaviour on the agents’ believability. The research is ultimately motivated to create agents that would naturally learn to exhibit different levels of altruism and be able to respond in a meaningful way to dynamic situations.

A selection of publications on this project are available here.

Altruism and Selfishness in Believable Game Agents: Deep Reinforcement Learning in Modified Dictator Games

Daylamani-Zad, D. and Angelides, M.C., 2020. Altruism and Selfishness in Believable Game Agents: Deep Reinforcement Learning in Modified Dictator Games. IEEE Transactions on Games.