Empowering consumers: an experimental study of human and AI intermediary in insurance decision-making

Insurance institutions are increasingly leveraging AI to optimize operations, and insurance intermediaries, which are designed to facilitate consumer decision-making, have emerged as a key area of AI adoption. Despite its potential, challenges such as consumer trust, acceptance of algorithmic decision-making, and ethical considerations raise questions about how AI will shape the role of intermediaries in influencing insurance decisions. This study investigates the impact of insurance intermediaries (human vs. AI) on insurance purchasing behaviors through an intertemporal consumption decision experiment involving the option to purchase insurance for unexpected expenses. By varying the availability of insurance and types of intermediaries across experimental treatments, we first establish two fundamental findings: (1) insurance smooths consumption and enhances lifetime utility, and (2) both human and AI intermediaries significantly promote insurance uptake. Contrary to our expectations, the experimental results reveal no overall difference in effectiveness between human and AI intermediaries. However, a heterogeneity analysis using causal tree algorithms highlights critical nuances: individuals with higher risk aversion exhibit a stronger trust in human intermediaries, leading to higher insurance purchase rates, whereas individuals with lower risk aversion show no significant trust differences between human and AI intermediaries. These findings provide actionable insights for insurance companies, emphasizing the need for strategies tailored to high-risk-averse consumers' preference for human guidance, while leveraging AI's potential to effectively engage low-risk-averse individuals. This study contributes to understanding the interplay between AI, trust, and consumer behavior, offering valuable implications for the design of AI-powered insurance services.