How do humans learn about other persons? How do we make decisions about our social behaviour towards them? These are key questions in social decision making research. Laboratory studies usually address passive forms of learning (e.g., trial-and-error learning). However, insights from artificial intelligence demonstrate that learning in real-life is less passive and much more self-directed. Humans select specific inquiries to acquire specific, relevant information (active learning). Importantly, there are social costs of information gathering to consider (e.g., too extensive question asking can be offensive, or time consuming). Understanding how people learn about others and make social decisions, requires understanding how people: 1. prioritize inquiries depending on which information is most relevant. 2. balance the benefit of gathering more information against the social costs. I propose a series of studies that combine methods from behavioral economics with computational models from artificial intelligence. This allows for the study of active social information gathering and the role of social costs. Project 1 and 2 examine multiple social information gathering costs. Project 3 systematically compares active with passive social learning. This novel, integrative approach will provide a mechanistic understanding of active social learning.