Although scholars have long studied knowledge search from a problem-solving perspective, research examining how technological changes shape the specific components of knowledge search, namely problem formulation and solution finding, remains scant. This study bridges this gap by teasing apart these components and examining how Artificial Intelligence-Generated Content (AIGC) technologies impact the solution finding process. Leveraging the advent of ChatGPT in a quasi-experiment design, our analysis of data from Stack Overflow, a platform for crowdsourcing coding knowledge, shows that the likelihood of programmers getting responses increases post-ChatGPT, after controlling for changes in problem formulation. However, this increase does not lead to increasing likelihood of locating satisfactory solutions, and instead results in a prolonged duration for locating accepted solutions as well as extensive post-acceptance discussions. Additionally, we also explore how the direction of progammers’ knowledge search moderates these effects and we attribute such heterogeneity to varying capabilities of solution evaluation, an overlooked conceptual factor in prior research.
