Two AI agents, Albert and Kai, are trained to play Tag using Deep Reinforcement Learning. Their actions are controlled by Neural Networks that are updated to maximize rewards and minimize punishments over time.