Principles on how to manage interactions between human workers and artificial intelligence/machine learning technologies

Causal ambiguity is a key issue when integrating artificial intelligence (AI) and machine learning (ML) technologies with human work activities. It reduces accountability, inhibits learning, and fuels uncertainty of the human workers and technology designers. We identify challenges and develop design principles to manage the human-machine interactions and reduce causal ambiguity. Our approach builds on three steps. First, we compare AI/ML technology to “traditional” automation technology and outline two characteristics creating causal ambiguity: interdependencies among the technologies and learning by the technology. Second, we look at human-machine collaborations as a form of principal-agent relationships and specify how humans interact with AI/ML technology by outlining three relational actions: information provision, performance evaluation, and monitoring. Third, bringing together the two characteristics and the three relational actions, we identify challenges of the human-machine collaborations and develop six design principles for managing the challenges.