The field of chemical engineering is witnessing a significant shift towards the integration of physical constraints into machine learning models. This trend is driven by the need for more reliable, generalizable, and interpretable models that can handle complex, multi-physics phenomena. Researchers are exploring innovative approaches to embed physical knowledge into machine learning frameworks, such as using physics knowledge graphs and direct preference optimization. These methods have shown promising results in reducing constraint violations and improving the accuracy of physically relevant parameters. Noteworthy papers in this area include: PKG-DPO, which introduces a novel framework that integrates physics knowledge graphs with direct preference optimization to enforce physical validity in AI-generated outputs. Physics-Constrained Machine Learning for Chemical Engineering, which summarizes recent developments and highlights challenges and opportunities in applying physics-constrained machine learning to chemical engineering.