Carl Westin

Summary v bility. While technology advancements have made it possible to increasingly tai- lor automation to the individual’s preferences, needs, and abilities, there are sev- eral technical challenges to be overcome before truly conformal automation can be developed, most importantly how to extract individual’s unique preferences and problem-solving style. Research is needed to establish the consistent and critical control parameters that would characterize a person’s problem-solving style. Automation designers need to carefully consider the goals and objectives in a specific domain for which a conformal system is considered. Many work domains, such as the flight deck, may be more suitable for facilitating homogeneity that re- stricts individual differences in interaction and problem-solving. Complementary to conformal automation providing explicit advisories, interface design should support the variability in problem-solving styles. Ecological interface design, for example, can be used to facilitate such personalized problem-solving by visualizing the “ob- jective truth.” That is, by showing the constraints affecting a situation, the operator is allowed to solve problems in her or his preferred way. Returning to the artificial intelligence (AI) driven personal assistants. Although still in their infancy in reality (e.g., Siri, Cortana, and Assistant), as they mature they are expected to significantly influence human-automation interaction to become a more fluid cooperation similar to the futuristic glimpses of automated characters observed in fiction (e.g., TARS, Samantha, and HAL 9000). While the acceptance of these systems depends on many aspects, the compatibility with the human, not the least their conformance, may have a significant effect on how willing people will be to interact, accept, and trust in it and its advice. While the extent to which acceptance and trust will depend on the intelligent assistant’s strategic conformance remains to be studied, this thesis indicates benefits thereof, at least in the context of ATC conflict solving. The temptation to personalize automation simply because it can be, however, should be avoided. Particularly in safety-critical domains, both benefits and drawbacks of such capable automation must be considered and evalu- ated before implementation.