The purpose of this project was to create an emergent flocking boid-based crowd behavior, through generalization of the system rules, predatory threat and foraging for the virtual crowd. Extensive research was conducted on the topics of flocking boids, prey and predator behaviors in nature and emergent behaviors and self-organizing systems. 3D Reinforcement learning environment was implemented using Unity ML-agents. Finally, evaluation was conducted by comparing two different ML crowd implementations with a flocking boid control environment. The results showed a successful evolving of an emergent crowd behavior from one of the models, using a generalized set of system rules. Despite the suffered hardware limitations, the exhibited results satisfied the problem criteria, as well as providing interesting insights and findings regarding emergent behaviors.