EXTENDED ABSTRACT: With the rise of data-driven methods as the fourth scientific paradigm, it has significantly impacted on the third scientiffc paradigm named physical model driven in the ffeld of alloy design. However, neither of the two paradigms can break the mutually exclusive relationship between model accuracy and interpretability, thus failing to meet the high efffciency and rationality requirements for alloy design in the ffeld of Material Genome Engineering. This dilemma has also given rise to the emergence of the fffth paradigm named AI4Sci in the ffeld of alloy design. This report provides an overview of various cases using the physical metallurgy guided artiffcial intelligence method system. It systematically explains how to deeply integrate physical models/mechanisms with artiffcial intelligence from three levels: numerical data guidance, image data guidance, and mechanism guidance, in order to break the mutually exclusive relationship between accuracy and interpretability. With the continuous deepening of the introduction of physical metallurgy principles, the artiffcial intelligence strategy optimization methods used in the entire model system are also constantly improving, and the amount of data required for model training is constantly decreasing. Under this framework, there is a synergistic deepening relationship between physical models/mechanisms and artiffcial intelligence strategies, making it easier to break the mutually exclusive relationship between model accuracy and interpretability. This report systematically discusses these three levels, fully exploring the deep integration of physical models and artiffcial intelligence models, the application environment and essential reasons of various technical solutions. Thus, it reveals the theoretical essence, advantages and disadvantages of the three paradigms named physical models, artiffcial intelligence, and AI4Sci. For cross scale modeling, materials science large model, etc., it provides guidance on the idea and technical methods for the future development of each scientiffc paradigm in the ffeld of alloy design.

Keyword: Alloy design; Material genome engineering; Artiffcial intelligence; AI4Sci.