The primary objective of this article is to adjust readers' expectations. Despite the current seemingly boundless optimism surrounding AI, its development is still subject to significant limitations. The challenges of AI transformation lie in technical difficulties, such as insufficient training data for models and alignment issues. Additionally, social and economic barriers will also restrict the impact of AI, necessitating support from social systems and norms. Therefore, it is essential to focus on innovations across various fields rather than over-relying on progress in a single area.