AI provides a new pathway to accelerate understanding and discover new materials that address global challenges across many sectors including the energy sector (e.g., battery technology), transportation (e.g., lighter weight aerospace materials), and health care (e.g., drug delivery devices, soft-tissue imaging, and implant materials) and many more areas.
Materials Science/Engineering/Chemistry/Physics majors:
In light of the Materials Genome Initiative, data science, and AI are being pursued to tackle problems in materials science across all materials domains, and are a key part of both developing a new understanding of underlying materials physics as well as developing robust methods for design of new materials. A few examples of open problems include:
- How to develop high-dimensional generative models for material behaviors.
- How to explore and use the underlying structure in data to develop invertible reduced-order models, suitable for computational analysis.
- How to leverage AI to improve efficiency and sustainability of material design by predicting the best next experimental sample/test conditions.
- How to use AI to bridge data and models across length and time scales to leverage quantum or atomistic features in predictions/understanding of macro-scale material response. Solutions to this type of scaling issue could have implications across all science.
Materials Science is a rich scientific domain in which to develop and deploy novel AI algorithms and strategies. Materials science is rich in datasets that are image-based (e.g., many forms of micrographs of material structure), model-based (e.g., physics-based predictive modeling codes at all length scales), and with wide-ranging feature sets (e.g., processing parameters, environmental conditions, test parameters). Opportunities to design innovative applications of AI/DS/ML/modeling that address materials challenges such as:
- How to create an interpretable neural network that tells you what patterns it's using to classify materials.
- How to design efficient optimization techniques to search through a large combinatorial space of materials.
- How to build physics into AI algorithms in order to better model a materials system.