Examining the Impact of Generative AI on Users' Voluntary Knowledge Contribution: Evidence from A Natural Experiment on Stack Overflow

• 成果类型: Article
• 署名作者: Shan, Guohou; Qiu, Liangfei
• 署名单位: Northeastern University; State University System of Florida; University of Florida
• 刊物名称: INFORMATION SYSTEMS RESEARCH
• ISSN/ISSBN: 1047-7047; 1526-5536
• DOI: 10.1287/isre.2023.0332
• 发表日期: 2026-06
• 关键词: knowledge contribution; generative AI; voluntary Q&A platforms; answer quantity; answer quality; difference-in-differences; communities; usage; difference; DESIGN; time

摘要

Voluntary knowledge contribution on online platforms holds significant value for users, platforms, and firms. Rapid advancements in generative artificial intelligence (AI) techniques have facilitated the automatic generation of knowledge on question-andanswer (Q&A) platforms. However, the impact of generative AI on users' voluntary knowledge contributions remains an empirical question. On the one hand, users may learn from generative AI, improving their answers by providing more organized and logical responses. On the other hand, generative AI can produce fabricated answers, and the accelerated pace of responding with AI assistance may impose additional cognitive burdens for comprehending the outputs, potentially reducing overall contributions. Our study examines the effects of generative AI, specifically ChatGPT, on users' voluntary knowledge contributions on Stack Overflow, one of the largest Q&A platforms. Utilizing a natural experiment, we employ difference-in-differences (DID) estimation to investigate the effects of generative AI on both the quantity and quality of user contributions, measured by the number of answers generated per day, answer length, and readability. Our findings reveal that the use of generative AI correlates with an increased number of answers generated by users, and these answers tend to be shorter in length and easier to read. We further explore the moderating effects of cumulative usage and usage intensity on the impacts of generative AI to test the mechanisms of learning and cognitive load. Our results indicate that users are learning from generative AI, enabling them to answer more questions while producing shorter and more digestible responses. Conversely, the additional cognitive burden associated with intensive AI usage negatively affects its impact on answer quantity. The implications of this study are both theoretical and practical. Theoretically, we contribute to the Information Systems (IS) literature by examining the influence of generative AI on users' voluntary knowledge contributions within the context of Q&A platforms. Practically, our findings provide platform owners and managers with insights into how generative AI affects users' knowledge contribution behavior, guiding decision-making and for AI into their