Harnessing or Hindering Human Cognition: Multi-Level Effects of AI-Augmented Decision Making in Global E-Commerce

The rapid advancement of artificial intelligence (AI), particularly large language models (LLMs) and intelligent agents, has profoundly reshaped how individuals and firms make strategic decisions. Among all sectors, cross-border e-commerce has become one of the most dynamic and AI-intensive domains. Platforms such as Alibaba International, Amazon, and TikTok Shop increasingly deploy AI-driven tools to assist sellers in product recommendation, trend analysis, translation, and market forecasting. These technologies have significantly improved operational efficiency and global market reach. However, growing reliance on AI has also raised concerns about its long-term effects on human cognition, learning, and independent decision-making. This project aims to examine how AI influences human cognition and learning in the context of cross-border e-commerce decision-making. At the individual level, it explores whether AI serves as a cognitive catalyst that stimulates reasoning and learning, or as a cognitive substitute that encourages cognitive offloading and metacognitive laziness. Overreliance on AI may reduce users’ ability to think critically and reflect on their own decisions, whereas collaborative engagement with AI may strengthen higher-order thinking and adaptive capacity. At the system level, the project investigates whether prolonged human–AI interaction can lead to convergence in cognitive and decision-making patterns across individuals, creating what we term “cognitive convergence.” It further examines whether dynamic system design—such as periodically varying the perceived role of AI—can reactivate reflective thinking and sustain cognitive diversity and adaptability within the system. Methodologically, the project employs a multi-method research design combining a randomized field experiment with a cross-border e-commerce platform (Xingyun), a controlled laboratory experiment using a self-built simulation platform (Glotra), and agent-based modeling (ABM) to capture emergent collective dynamics. These studies jointly trace how individual cognitive processes evolve and aggregate into system-level patterns of convergence and adaptation. By linking micro-level cognition with system-level evolution, this project provides theoretical and practical insights into how AI can be designed to enhance rather than replace human intelligence. The findings will offer guidance for platform developers and policymakers seeking to promote sustainable, human-centered, and cognitively diverse AI ecosystems in the digital economy.