Materials science and engineering deals with materials having different compositions, going through different manufacturing processes resulting in different microstructural arrangements that affect the material properties and their performance in an application. Understanding the correlations between composition, processing parameters, structure and properties of a material (so that one can alter the properties to meet specific performance requirements) is at the core of materials science. A huge design space needs to be systematically explored to design a material (and the required processing) with target properties. Traditionally, materials scientists have used a combination of intuition, experimentation and numerical simulations to address this problem. Sometimes, optimization around response surfaces built from the data is also used to make the task tractable. The experimentation and simulation generates a large amount of data when multiple such situations are combined but, it is generally not harnessed together systematically. As noted in the 2021 strategic plan of the Materials Genome Initiative, AI/ML techniques can potentially transform this entire landscape of materials research by providing predictive capabilities where traditional physics-based models either do not yet exist or are too slow and computationally expensive to be useful for systematic exploration of the design space. The focus of this workshop is at the intersection of artificial intelligence, machine learning and materials science.

Despite a surge in publications on AI-enabled materials discovery within materials science literature in the last five years, the interaction between machine learning researchers and materials scientists (especially, scientists working on structural materials, their microstructures, textures and so on) has been very sparse. On the other hand, AI/ML techniques can clearly be integrated into materials design frameworks (e.g., MGI efforts) to support accelerated materials development, novel simulation methodologies and advanced data analytics. Hence, there is an immediate need for exchange of ideas and collaborations between machine learning and materials science communities. Moreover, to realize the true potential of AI/ML for materials research, the materials' data needs to be interoperable and reusable so that the models can be applied on disparate data sets (or material systems). However, various sources of data (in different standards) and often sparse nature of materials data have been great challenges for practitioners. In fact, achieving “AI-ready” data has been identified as one of the goals of the MGI effort (MGI Strategic Plan 2021). Toward this goal, spreading awareness about “AI-ready” data policies among materials researchers and building tools to assess quality of data will go a long way. This can be achieved through deeper collaborations between the two fields. The aim of this workshop is to bring together these two communities and foster deeper collaborations to accelerate the adoption of AI/ML in materials science. Finally, the exchange of ideas and collaborations will also help the AI/ML community to discover several challenging problems which are essential for solving problems in materials science; for example, the problem of incorporating domain knowledge to improve data efficiency of deep neural networks (so that they can be used as an alternative to computationally expensive simulation models in materials science).