This study explores the impact of ChatGPT on Stack Exchange's sustainability and community health by analyzing data from 40 communities from September 2021 to September 2023. In the community-level analysis, we used the difference-in-differences method, utilizing the previous year's data to construct the control group. We found the launch of ChatGPT significantly decreased the number of new posts on the platform, with the rollout of GPT-4 intensifying this effect. Additionally, our findings indicate notable heterogeneity in the impacts of GPT across various communities. Technology communities experienced the most significant negative effects. In the user-level analysis, we found the reduction in post numbers primarily resulted from decreased activity among users registered in the last two years. Subsequently, we used the Temporal Causal Inference (Turjeman and Feinberg, 2023) method using data from users who registered 1-6 months earlier to construct control groups and conducted regressions using the Difference-in-Differences and Generalized Synthetic Control Method. We identified the effect of GPT by comparing the number of posts from users with different registration times during the same registered duration. We found the negative GPT effect consistent with previous results, and it is more pronounced among users with higher community status and more social interaction.