Search for: All records

Artificial Intelligence (AI) and Machine Learning (ML) capabilities have the potential for large-scale impact to tackle some of the world’s most pressing humanitarian challenges and help alleviate the suffering of millions of people. Although AI and ML systems have been leveraged and deployed by many humanitarian organizations, it remains unclear which factors contributed to their successful implementation and adoption. In this study, we aim to understand what it takes to deploy AI and ML capabilities successfully within the humanitarian ecosystem and identify challenges to be overcome. This preliminary research examines the deployment and application of an ML model developed by the Danish Refugee Council (DRC) for predicting forced displacement. We use qualitative methods to identify key barriers and enablers from a variety of sources describing the deployment of their Foresight model, a machine learning-based predictive tool. These results can help the humanitarian community to better understand enablers and barriers for deploying and scaling up AI and ML solutions. We hope this paper can spark discussions about the successful deployments of AI and ML capabilities and encourage sharing of best practices by the humanitarian community. 

Abstract Standardized design approaches such as those embodied by concurrent design facilities have many benefits, such as increased efficiency of the design process, but may also have hidden costs. Specifically, when their standardized organizational decomposition is a poor fit for the particular design problem, important design trades might be missed or poor decisions made. Before we can understand how this lack of fit impacts the design process, we must be able to empirically observe and measure it. To that end, this paper identifies measures of “fit” from the literature along with attributes likely to impact design process performance, then evaluates the measures to determine how well the measures can detect and diagnose potential issues. The results provide comparative insights into the capabilities of existing fit measures, and also build guidance for how the systems engineering and design community can use insights from the “fit” literature to inform process improvement. 

The socio-technical perspective on engineering system design emphasizes the mutual dynamics between interdisciplinary interactions and system design outcomes. How different disciplines interact with each other depends on technical factors such as design interdependence and system performance. On the other hand, the design outcomes are influenced by social factors such as the frequency of interactions and their distribution. Understanding this co-evolution can lead to not only better behavioral insights, but also efficient communication pathways. In this context, we investigate how to quantify the temporal influences of social and technical factors on interdisciplinary interactions and their influence on system performance. We present a stochastic network-behavior dynamics model that quantifies the design interdependence, discipline-specific interaction decisions, the evolution of system performance, as well as their mutual dynamics. We employ two datasets, one of student subjects designing an automotive engine and the other of NASA engineers designing a spacecraft. Then, we apply statistical Bayesian inference to estimate model parameters and compare insights across the two datasets. The results indicate that design interdependence and social network statistics both have strong positive effects on interdisciplinary interactions for the expert and student subjects alike. For the student subjects, an additional modulating effect of system performance on interactions is observed. Inversely, the total number of interactions, irrespective of their discipline-wise distribution, has a weak but statistically significant positive effect on system performance in both cases. However, excessive interactions mirrored with design interdependence and inflexible design space exploration reduce system performance. These insights support the case for open organizational boundaries as a way for increasing interactions and improving system performance. 

Abstract The engineering of complex systems involves large number of individuals within multiple organizations spanning multiple years. Since it is challenging to perform empirical studies directly on real organizations at scale, some researchers in systems engineering and design have begun relying on abstracted model worlds that aim to be representative of the reference socio-technical system, but only preserve some aspects of it. However, there is a lack of corresponding knowledge on how to design representative model worlds for socio-technical research. Our objective is to create such knowledge through a reflective case study of the development of a model world. This “inner” study examines how two factors influence interdisciplinary communication during a concurrent design process. The reference real world system is a mission design laboratory (MDL) at NASA, and the model world is a simplified engine design problem in an undergraduate classroom environment. Our analysis focuses on the thought process followed, the key model world design decisions made, and a critical assessment of the extent to which communication phenomena in the model world (engine experiment) are representative of the real world (NASA's MDL). We find that the engine experiment preserves some but not all of the communication patterns of interest, and we present case-specific lessons learned for achieving and increasing representativeness in this type of study. More generally, we find that representativeness depends not on matching subjects, tasks, and context separately, but rather on the behavior that results from the interactions of these three dimensions.