Behavioral mechanisms underlying animal movements include social interactions among individuals comprising the group. The increasing relevance of machine learning in theoretical biology has made it possible to investigate the intricate interplay between collective traits and locomotor and cognitive properties of individuals. Our research explores the influence of social cues on the behavior of an intelligent agent within a school of fish, utilizing the smart self-propelled particle framework. In this agent-based model framework, agents are imbued with an artificial neural network that determines their movements. The school is governed by basic swarming rules such as alignment of individuals to the direction of nearby fish and attraction to the center of mass. Our aim is to identify what social cues are relevant for an intelligent agent to adhere to the collective movement. We employ a selective pressure methodology, identifying the individuals that have been the most successful at responding to social cues. To validate our simulation predictions, we observe whether these trained agents can autonomously form new swarms that exhibit cohesive swarming behavior. By selecting the top-performing agents, we aim to pinpoint the optimal values and behaviors that support successful integration into the swarm. As opposed to previous research utilizing models inspired by physics, employing artificial neural networks within this framework can pave the way for future scientific endeavors to investigate the constraints that lead to observed behaviors combining movement and decision-making.