Human decision making in AI augmented systems: Evidence from the initial coin offering market

Abstract

The growing consensus that human intelligence and artificial intelligence are complementary has led to Human-AI hybrid systems. As digital platforms incorporating human-AI hybrids, platform designers need to evaluate the influence of AI on human judgment, and how such hybrid systems perform. In this paper, we investigate: Are human decisions influenced by AI agents in high uncertainty environments, such as evaluating ICO projects? Under what situations are humans able to mitigate AI agents-induced errors? Our results suggest that in general, humans are influenced by AI agents. Humans tend to use AI as a filter to rule out low quality projects, while a high AI rating triggers human expert to apply their own judgment.