Examining the Impact of Platform-initiated Generative AI Answers on Users’ Voluntary Knowledge Contribution Using a Natural Experiment 

Yuan Dong, Guohou Shan. 

Abstract: Platform owners are increasingly embedding generative AI directly into question-and-answer (Q&A) communities, yet the consequences of platform-initiated AI answers for subsequent human contribution remain poorly understood. This issue is important because platform AI does not merely compete with community members from outside the platform; it is displayed inside the focal community, benefits from privileged interface placement, and may reshape both learning and attention allocation. In this study, we examine how exposure to platform-initiated generative AI answers affects users’ subsequent voluntary knowledge contribution on SegmentFault, a large Chinese technical Q&A platform. Leveraging a natural experiment in which a subset of questions received AI-generated answers, we construct a user-level natural experiment and estimate the downstream effects of users’ first exposure to platform AI answers on later questioning and answering behavior using a difference-in-differences (DID) estimation. Grounded in social learning theory and the attention-based view, we theorize that platform AI answers serve simultaneously as public demonstrations from which users can learn and as salient attention-directing artifacts that reallocate community attention. We find that after users are exposed to AI answers, they generate a larger number of questions, attach more tags to each question, receive fewer likes per question, generate more answers, and attain a higher answer acceptance ratio. Moreover, user tenure strengthens the positive effect of AI exposure on the number of questions generated while mitigating the negative effect on likes received per question. These findings contribute to research on generative AI, online knowledge contribution, and platform design by showing that internal AI answerers can stimulate human contribution volume while simultaneously altering the visibility and recognition of that contribution. The results also offer guidance for platforms seeking to balance AI assistance with the health of human knowledge ecosystems.

Keywords: generative AI, online Q&A communities, platform AI, knowledge contribution, social learning, attention-based view, natural experiment