How do individual information-seeking preferences affect collective problem-solving in diverse settings? We often choose whom we collaborate with or seek information from. Self-selection can be driven by proclivities towards perceived merit or a preference for those who offer a different perspective. Yet our preferences can have profound systemic outcomes, even if opportunities abound to interact with diversity. We build upon the extensive tradition of collective problem-solving using agent-based modeling (ABM) to test this. We populate communicative networks of diverse problem-solving agents tasked with solving a complex problem too difficult to do alone. Agents can either exploit their neighbors’ solutions or explore for a new solution using their unique and diverse problem-solving ability. However, agents are also allowed to seek out new ties from the network. We test three conditions, where diverse agents in the network harbor proclivities towards (1) diversity (different types of neighbors), (2) homophily (same type of neighbors), or (3) merit (the current performance of their neighbors irrespective of type). We find that diversity-seeking not only leads to higher quality solutions, but also allows for these solutions to better disseminate to the rest of the network.