Cross-border data flows and AI adoption: Agent-based model simulations

Abstract

This paper develops a dynamic Agent Based Model to study the role of cross-border data flows for the joint uptake of artificial intelligence enabled software in manufacturing and engineering. The model features two technology-related business models: engineering as a face-to-face consultancy service, and engineering as a software licensing service. Engineering agents harvest data from their software clients in the home country and abroad and use the data for quality assurance and software updates. We compare scenarios along two dimensions: (i) harvesting data from own clients only versus from open data repositories, (ii) the strength of competition measured by the probability that a contract will be extended by another period. We find that restrictions on cross-border data flows slow down the speed of adoption considerably, particularly in small countries. The simulations generate an S-shaped technology uptake path for manufacturers and a U-shaped relationship between competition and technology uptake in engineering. Interestingly, cross-border data flows flatten the U.