Carl Westin

3-2 Automation acceptance in ATC 43 Interestingly, from this perspective there is little difference between trust as it de- velops in a machine versus a human agent. Notice that the rejection of automation may derive from the notion of dispo- sitional trust. That is: in the absence of experience, and before one can calibrate dynamic trust, how inclined is one to use automation? Is there already a prejudice against using automation? This issue was at the core of our research questions, as outlined in the following section. 3-2-3 Research questions Successfully introducing advanced ATC automation might rely heavily on controller acceptance, at least for some transitional period. It was therefore reasonable to hypothesize that controller acceptance would rely on the machine working in a way that was familiar to the human. We captured this notion in the concept of strategic conformance, which we defined as the degree to which automation’s behavior and apparent underlying operations match those of the human . We then set out to explore how usage and acceptance would be impacted by the possibly interactive effects of three factors: • Strategic conformance, • traffic complexity, and • level of automation (LOA). The main aim of the MUFASA project was to investigate the possibility that controllers would show a systematic bias against automation, which could jeop- ardize the introduction of advanced forms of ATM automation. That is, would controllers be accepting of automation that is designed to replace aspects of their strategic decision-making in the areas of conflict detection and resolution? Specific research questions included the following: • Conformance and acceptance - Are controllers more likely to accept auto- mated advisories when these mimic the controller’s own solution? • Complexity and acceptance - Everything else being equal, does acceptance vary by air traffic complexity? • Conformance and other effects - Do measures such as difficulty rating or re- sponse time show an impact of strategic conformance? These questions draw inspiration in part from the CORA project 35 which, again, tried to build automation capable of solving problems like a human would. Notice