Protein structure prediction methods, exemplified by AlphaFold, have achieved near-experimental accuracy in predicting protein structures. AlphaFold has been widely applied in scientific fields, such as life sciences, synthetic biology, drug discovery, and materials design. This article provides a systematic overview of the major advances in protein structure prediction, focusing on the technical evolution of AlphaFold and its limitations in predicting dynamic conformational changes, complex interactions, and conformations under non-native conditions. Based on this analysis, key challenges in the current field are identified, and potential strategies to address these issues are proposed. The review offers perspectives on future directions for improving protein structure prediction research, providing insights for subsequent methodological enhancements and broader applications.