With the rapid development of artificial intelligence, the application of machine learning algorithms in the steelmaking industry has become a research hotspot. This paper systematically explores the challenges and opportunities of intelligent models in complex industrial scenarios of short-process steelmaking, with a focus on analysing their current applications in key stages such as electric arc furnaces, refining, and continuous casting. An examination of typical scenarios in the steelmaking process elaborates on the role of machine learning in process optimization, anomaly detection, and autonomous decision-making. In response to the real-time and reliability demands of the steelmaking environment, this study proposes research directions for machine learning within intelligent manufacturing systems, including cutting-edge technologies such as multimodal sensing, causal reasoning, and digital twins. Finally, this study explores the challenges, potential solutions, and future application prospects of machine learning in deep integration with the short-process steelmaking industry.