Paper short abstract:

In a context where AI systems are pushing the aviation industry to interrogate the value of its technology in relation to humans, this paper describes how pilots interpreted and evaluated AI assistants—and the kind of assistance they wanted from them.

Paper long abstract:

As AI systems push the aviation industry to interrogate the value of its technology in relation to humans, this paper uses a critical co-design approach to describe how pilots interpreted and evaluated AI assistants. Pilots who participated in our research saw the digital flight assistants as teammates who would be generally untrustworthy, but, under the right circumstances, very helpful. This evaluation relied upon comparing the DFA with human co-pilots alongside three specific kinds of labour experiences where AI would fail to help them cope with increasingly degraded working environments. Yet, this mistrust was also couched in a layer of optimism. Paradoxically, in a future that they saw as necessarily getting rid of (most of) the humans in the cockpit, DFAs were imagined as possible conversational partners, reminiscent of the golden age of aviation. Taken together, these two aspects suggest the need for an approach to the evaluation of AI that combines structural and emergent factors--blending insights from design anthropology and critical political economy.