Symbiotic Co-Evolution in Collaborative Human-Machine Decision Making: Exploration of a Multi-Year Design Science Research Project in the Air Cargo Industry

Abstract

The work world is set to undergo major changes thanks to advancements in automation and artificial intelligence and is beginning to promote new forms of collaboration. The transition from a technologysupporting environment to a collaborative environment in which people and technology work together to achieve their goals requires a fundamental change in the way we design, build, and ultimately deploy information systems. Most work on information system design focuses on the effective augmentation of humans. However, little is known about constructing a sustainable mutually beneficial collaboration between human and machine. To better understand this relationship, we perform a case study drawing on ethnographic evidence collected during a multi-year design science research project with a major service provider for unit load device management in the air cargo industry that resulted in an artifact for human-machine collaboration (HMC). Our study takes a closer look at the co-evolution that emerges from the collaboration between human and artificial agents over time, in which both parties influence each other, the underlying tasks, and their environment. Our analysis reveals three facets of symbiotic co-evolution: agents' evolution, activity evolution, and structural evolution. The findings contribute to the HMC knowledge base and have implications for future HMC design initiatives.

Becoming Team Members: Identifying Interaction Patterns of Mutual Adaptation for Human-Robot Co-Learning

“…If we look at human-robot interaction literature, several terms are used that describe a similar process in which two parties or systems change their behavior and/or mental states concurrently while interacting with each other. Co-adaptation (Xu et al, 2012;Chauncey et al, 2016;Nikolaidis et al, 2017a) and co-learning (Bosch et al, 2019) are two of them, but we also encounter coevolution (Döppner et al, 2019), in which "co" stands for collaborative, also meaning "mutual. "…”

Theoretical Underpinnings and Practical Challenges of Crowdsourcing as a Mechanism for Academic Study

“…That is, we need to understand how people plan, coordinate and manage their collective actions as stable social structures in the context of large-scale distributed work, more precisely 'crowd work' [9]. From a social analytic and system development position, a co-evolving relationship between hybrid human-machine interactions [48] constitutes a new class of socio-technical systems.…”

“…Human crowds are not an exception. The nature's ecosystem is a highly complex adaptive, selfregulating system where living organisms interact within the environment to sustain and even adapt to its changes [48]. That is, the biological properties of individuals influence the way as everything change and organisms' observation is an important part of sociotechnical research.…”

“…A large and growing body of literature (e.g., [27,48]) has attempted to investigate the socio-technical outcomes of collective hybrid intelligence. A key challenge for a system integrating AI and crowdsourcing for scientific purposes relies on modeling intelligent behavior [30] while providing support for vast amounts of users with diverse background and dispersed geographically contributing over multidimensional data fields.…”

“…That is, thousands of unconnected outputs can be aggregated in order to train algorithms for providing better decisions [54]. In a similar way, the crowd becomes more skilled and accurate as a result of this co-evolving relationship [9,48]. Consistent with this suggestion, Dong et al [33] came up with more or less the same aspects by arguing that the integration of crowd inputs into AI systems requires a comprehensible translation of system states and requirements to humans to elicit high quality outputs from the crowd.…”