Abstract

As artificial-intelligence-assisted tools become prevalent in supply chain inventory management, algorithmic trust critically determines the success of human-AI interaction. Based on the newsvendor decision making, this study primarily considers two key factors—information construal level and presentation format—to investigate how they affect trust in AI recommendation, alongside the moderating role of decision-makers' individual traits. Through an incentivized 2×3 between-subjects experiment, we empirically analyze the level of trust in AI decision-making tools, measured by participants' Mean Absolute Percentage Deviation (MAPD) from the AI-advised values across different groups. Results indicate that text-and-graphic and scenario-based presentation formats significantly reduce decision deviation and enhance AI trust, whereas the construal level shows no significant impact. Furthermore, while decision-making style significantly influences AI tool utilization frequency, it does not affect the actual adoption of AI advice. This study clarifies the mechanisms driving algorithmic trust in newsvendor settings and provides empirical evidence and guidelines for enterprises to optimize the front-end interactive design of AI decision assistants and implement differentiated promotion and literacy training strategies tailored to varying employee traits.

Download

Share

COinS