Finding the Path Toward Design of Synergistic Human-Centric Complex Systems

Modern decision support systems are becoming increasingly sophisticated due to the unprecedented volume of data that must be processed through their underlying information architectures. As advances are made in artificial intelligence and machine learning (AI/ML), a natural expectation would be to assume that the complexity and sophistication of these systems will become daunting in terms of comprehending their design complexity, effective operations, and managing total lifecycle costs. Considering the fact that such systems operate holistically with humans, the interdependencies created between the information architectures, AI/ML processes and humans begs that a fundamental question be asked –“how do we design complex systems such as to yield and exploit effective and efficient human-machine interdependencies and synergies?” A simple example of these interdependencies may include the effects of human actions changing the behavior of algorithms and vice-versa. The algorithms may serve in the extraction and fusion of heterogeneous data, employ a variety of AI/ML algorithms that range from hand crafted, supervised and unsupervised approaches coupled with federated models and simulations to reason and infer about future outcomes. The purpose of this chapter is to gain a high-level insight into such interdependencies by examining three interrelated topics that can be viewed as working in synergy towards the development of human-centric complex systems: Artificial Intelligence for Systems Engineering (AI4SE), Systems Engineering for Artificial Intelligence (SE4AI), and Human Centered Design (HCD) and Human Factors (HF). From the viewpoint of AI4SE, topics for consideration may include approaches for identifying the design parameters associated with a complex system to ensure code maintainability, to minimize unexpected system failures, and to ensure that the assumptions associated with the algorithms are consistent with the required input data while optimizing the appropriate level of interaction and feedback from the human. Considering SE4AI, how can the synergies between different AI/ML approaches from handcrafted rules to strictly data-driven learning within the data-to-decisions information pipeline be realized, again while maximally leveraging human inputs? From the lens of HCD and HF, a system is likely to be necessarily complex, and a key aspect of the designer may be to ensure an optimal balance between the human systems or software developer and the end-user. For instance, can principles from HCD/HF engineering permit us to design better systems that enhance end-users’ strengths (e.g., intuition, novel thinking) while helping to overcome their limitations (e.g.,