Balancing privacy considerations and customization preferences for consumer: LLMs adoption and coordination strategies in supply chains

Abstract

The manufacturer’s motivation to invest in product customization is reduced by consumer uncertainty regarding customized products and the double marginalization effect in supply chains. The emergence of large language models (LLMs), with their enhanced interaction and inference capabilities, offers opportunities to mitigate consumer uncertainty in customized product sales, but simultaneously raises novel privacy concerns. Consequently, supply chain members encounter operational decision complexities in the context of product customization. This study analytically examined a two-tier supply chain and investigated the adoption of LLMs in customized product sales under consumer uncertainty and privacy concerns. We further explored revenue-sharing and cost-sharing contract mechanisms for supply chain coordination. The results demonstrated that adopting LLMs with price discrimination generally benefited the manufacturer, the retailer, consumer surplus, and social welfare, while adopting LLMs without price discrimination only benefited those when privacy costs were low. The product-customization level was not affected by the adoption of LLMs without or with price discrimination in wholesale contracts. Both revenue-sharing and cost-sharing contracts enhanced manufacturers’ profits and increased product-customization level, while retailers benefited only when the revenue-sharing or cost-sharing ratio was low. Privacy costs hindered Pareto improvements only in the adoption of LLMs with price discrimination under revenue-sharing contracts. Finally, we characterized the conditions under which manufacturers and retailers reach contractual agreements regarding wholesale prices, cost-sharing, and revenue-sharing, and we explored the impact of privacy costs on such agreements.