Artificial Intelligence (“AI”) in recent years has developed from a novel concept to practical applications throughout the global economy. AI increasingly performs cognitive tasks and has evolved into the role of a human’s teammate. However, algorithms are not designed to facilitate a teaming process. Many firms do not adequately investigate the right combinations of the strengths of machines (computing power and memory storage) and humans (e.g. intuition and expertise) as well as the dynamics of their interactions, which undermines project performance.
To address the challenge of effective AI system design, MIT’s Digital Transformation Research Group developed the human-machine teaming (“HMT”) framework. It proposes that successful AI projects are enabled by appropriate configurations of HMT capabilities under the respective decision context of the AI project, which is characterized by the level of decision risk and AI’s decision-making process. Building on these efforts, this paper provides three core results: 1) expansion and validation of the conceptual HMT capability framework and empirical assessment of AI projects; 2) recommendation of a quantitative assessment instrument for future research; 3) provision of recommendations to supply chain leadership for successful AI implementations.
The authors refined the conceptual HMT framework and propose that AI project success can be explained by the effectiveness of human-machine Mutual Learning, which is enabled by the HMT capabilities of Transparency, Authority Balance and Secure Interaction. The authors empirically validated the HMT capability conceptual framework via multiple case study research methodology, assessing 22 case studies of AI applications in supply chain management and conducting in-depth semi-structured interviews with two companies. Six academic propositions were derived from the results. For example, it was shown that seven foundational HMT capability indicator concepts are required for successful supply chain AI project implementation and that decision context is the determinant of HMT capability configurations. As AI projects evolve, they change position within the decision-context framework and require different capabilities as learning occurs. Related managerial recommendations were derived and an assessment 3 tool for validation of the HMT capabilities’ structural relationships was developed. The validated framework serves as a guideline for supply chain professionals in AI project implementations and assessments.