Intense drone traffic, exceeding human capabilities of manual control, is expected to occur during the last stage of Unified Traffic Management (UTM) and Unmanned Airspace System (UAS) service deployment in cities. In this paper, we discuss how humans and automation could collaborate to manage this airspace. We review theory on options for UTM airspace structure (volumes, points, networks, layers), machine learning, optimization, and human-automation collaboration. Based on simulation and visualization of two cities, we discuss four abilities: to discern traffic patterns, to recognize situations, to predict situational developments, and to function in varying conditions of rule-following habits of airspace users. We then discuss the challenge of collaborating though the use of advanced visual dashboards, for human-in-the loop AI but also for society-in-the-loop. Finally, we discuss how the challenge of humanautomation collaboration can be expected to shift, as the capabilities of the machine increases.