Equilibrium analysis of live-streaming e-commerce supply chain in the AI era

Live-streaming e-commerce is an emerging online shopping channel that provides consumers with a satisfying shopping experience. This paper explores the operational and strategic implications of integrating artificial intelligence (AI) into live-streaming, focussing on the partial uncertainty elimination effect of AI avatar streamers and the full uncertainty elimination effect of real human streamers. We develop game-theoretical models to evaluate whether the retailer or the manufacturer should engage in AI and traditional live-stream selling under different selling modes, i.e. reselling mode and marketplace mode. We find that highly polarised AI costs, i.e. either extremely high or extremely low, and low human costs make it easier for the retailer to profit from AI live-stream selling under both selling modes. With high hassle costs, AI live-streaming's pricing advantage weakens, and the retailer may lower prices to enhance its competitiveness. Interestingly, implementing AI live-streaming changes how commissions impact the choice of the selling mode in the context of live-streaming. Our results provide several practical managerial implications for effectively integrating AI into live-stream selling, ensuring that AI integration delivers tangible benefits in real-world settings.