AI Inter-organizational research

As artificial intelligence (AI) increasingly mediates inter-organizational decision-making, concerns arise about how algorithmic opacity reshapes trust, information sharing, and cooperation between firms. This study investigates the relational consequences of AI-supported negotiation, focusing on explainability as a sociotechnical governance mechanism. Using a dyadic negotiation experiment involving 76 professional dyads across three global software providers, we compare human-only negotiation with opaque (black-box) AI and explainable AI (XAI) decision-support systems. Results show that black-box AI significantly reduces partner-directed trust, voluntary information sharing, and subjective satisfaction, despite improving negotiation efficiency. In contrast, explainable AI restores—and in some cases enhances—relational outcomes, yielding higher trust, earlier cooperative gestures, and greater willingness to disclose strategic information. Moderated mediation analyses reveal that explainability mitigates the negative indirect effect of AI on satisfaction through trust. These findings demonstrate that explainability is not merely a technical usability feature but a relational requirement that shapes how organizations interpret intentions and evaluate fairness in AI-mediated collaboration. By articulating explainability as a relational governance mechanism, this study advances theory on AI-enabled cooperation and offers practical guidance for the responsible design of negotiation and procurement systems in digitally mediated inter-organizational environments

随着人工智能（Artificial Intelligence，AI）日益成为组织间决策的中介载体，学界与业界均产生了关于算法不透明性如何重塑企业间信任、信息共享与合作关系的诸多担忧。本研究以可解释性作为社会技术治理机制，聚焦探究人工智能辅助谈判对组织间关系产生的影响。本研究开展一项双边谈判实验，招募来自三家全球软件供应商的76组专业配对受试者，对比纯人工谈判、不透明（黑箱）人工智能与可解释人工智能（Explainable AI，XAI）决策支持系统的应用效果。实验结果表明，尽管黑箱人工智能可有效提升谈判效率，却会显著降低合作方信任度、自愿信息共享意愿与主观满意度。与之形成鲜明反差的是，可解释人工智能不仅能够修复受损的关系绩效，在部分场景中甚至可优化关系产出，带来更高水平的信任、更早出现的合作意向行为以及更强的战略信息披露意愿。有调节的中介分析结果显示，可解释性可通过信任路径，削弱人工智能对满意度产生的负向间接影响。本研究证实，可解释性并非仅为一项技术可用性功能，而是关系层面的必要条件，其可影响组织在人工智能介导的协作中对合作意图的解读与公平性评估。本研究将可解释性明确为一种关系治理机制，既推进了人工智能驱动合作领域的理论发展，也为数字化介导的组织间环境下谈判与采购系统的负责任设计提供了实践指导。