Carl Westin

40 First empirical insights These questions are neither trivial nor merely academic. The MUFASA project assumed that future ATMwill rely increasingly on automation capable of overseeing the cognitive and strategic aspects of ATM. The SESAR target concept for future ATM 7 is built on an evolutionary path of five “Service Levels” that correspond to progressively more sophisticated automation, in terms of both the types of tasks it can perform and the level of authority and autonomy it can assume. 3-2 Automation acceptance in ATC Given the increasing need for automation to accommodate future traffic and com- plexity levels, how do we go about developing automation that controllers will ac- cept, trust and come to use? It has been noted that, in a variety of fields, operator acceptance might be one of the greatest future obstacles to successfully introducing new automation. 51 The ATM community has for years struggled to find a suitable automation design approach that promotes a functional and collaborative human- machine relationship. Past efforts such as AERA 74, 75 and ARC 2000 76 rested on a technology-centered approach that assumed optimized, algorithmic CD&R. The optimized automation view has softened somewhat in recent years, and as the movement toward “human-centered” automation design 51 has grown, so too has the recognition that optimized algorithmic approaches to CD&R are often at odds with the more heuristic approach controllers bring to problem solving. 30, 32, 33 Given advances in flight path monitoring and optimization, automated decision aids will likely introduce solutions that are: 14, 51, 127 • more strategic in timescale; • more strategic in their ability to consider greater airspace volume and network effects; • less transparent to the controller, in that all decision constraints are not made clear, and • presented as suggestions. Central to this discussion is the role of operator acceptance. Among the various risks in introducing new automation is the potential for lack of operator acceptance to limit the use and, in turn, the potential safety and performance benefits of such automation, 13, 51 both in ATC, 6, 128 and in other high reliability domains. 65 Notice one paradox of advanced automation: as we move toward more capable CD&R automation, intended benefits can only be realized if a system is properly used, yet such a system might only be used if its benefits are recognized. 129 Otherwise, a