Collaborative Intelligence: A Scoping Review Of Current Applications

Abstract

This review provides a novel examination of the emerging field of collaborative intelligence and demonstrates the value that human-AI teams can deliver. Humans and artificial intelligence (AI) systems have complementary strengths. This complementarity creates the potential to achieve a step-change in performance by combining inputs from human and AI on a common task. We introduce the construct of “collaborative intelligence” and develop a set of criteria, for evaluating whether an AI system enables collaborative intelligence. Applications utilizing collaborative intelligence had to have (1) complementarity (i.e. the collaboration draws upon complementary human and AI capability to improve outcomes), (2) a shared objective and outcome, and (3) sustained, two-way task-related interaction between human and AI. A systematic review of 1,250 AI applications published between 2012 and 2021 was carried out to investigate whether real-world examples of “collaborative intelligence” could be identified. The review yielded 16 AI systems which met the criteria, demonstrating that collaboration between humans and AI systems is possible and that these systems offer a wide range of performance benefits including efficiency, quality, creativity, safety, and human enjoyment.