Filling an AI and Materials Science Training Gap

The National Science Foundation has awarded Duke University a $3 million, five-year research traineeship grant to develop a program for graduate students to develop expertise in using artificial intelligence for materials science research.

The AI for Understanding and Designing Materials program, or aIM, will fill a vital workforce gap by training the next generation in the new convergent field of materials and computer science research.

“To achieve the promise of the Materials Genome Initiative of accelerated discovery, design and application of new materials, we must integrate the traditional tools of experimentation, theory and computation with the emerging tools of data science to transform the way we approach materials understanding and discovery,” said L. Catherine Brinson, chair of Mechanical Engineering & Materials Science at Duke and the director of aiM.

The aiM program at Duke will train 50 PhD students, funding 25 of them, from degree programs in computer science, data science, statistical science, and all materials disciplines including materials science, physics, chemistry, and all engineering fields.

The Materials Genome Initiative (MGI) is a multi-agency project of the United States Government to accelerate the development and deployment of new and advanced materials to address challenges related to clean energy, national security, and human health and welfare.

“The MGI promoted a paradigm shift from slow individual experiments and computation to the beginnings of data-driven AI approaches in materials science research,” added Brinson.

The research landscape in materials science has indeed begun to embrace these principles in ways such as high-throughput experimentation on novel compounds, data mining on arrays of computational predictions, and the use of machine learning (ML) to optimize composite constituents.

However, the field has not yet achieved the vision of training the next generation of materials experts to integrate AI into materials science and discovery.

One of the keys to success in developing a workforce in this exciting domain is recognizing its transdisciplinary nature, said Cynthia Rudin, associate director for research and mentoring of aiM, and professor of computer science, electrical and computer engineering, mathematics and statistical science.

“Students trained in materials science, who may be studying materials science, mechanical engineering, chemistry, biomedical engineering or physics, follow core curriculum requirements in those departments.,” said Rudin. “While a mathematics course is often part of the core requirements, neither a computational nor data science course is typically required in any major PhD program in the United States.”

Students pursuing research at the intersection of AI and materials science are left to individually find courses that may be relevant, yet they may be missing prerequisites for many data science courses and often have trouble enrolling due to high demand.

Following individual development plans designed for each student, the aiM program will integrate across all five years of each of their home department PhD degree programs through specialized coursework, professional skills development and experiential internships. Upon completion, each student will be awarded an aiM PhD Certificate of Specialization, in addition to their doctoral degree.

To capture and disseminate best practices and impacts on participating students, faculty and curricula, aiM will implement a comprehensive evaluation plan designed by Glenda Kelly, lead evaluator, in collaboration with aiM’s leadership team.

aiM Fellows will join their cohort of trainees each year in a summer boot camp, which will address fundamental skills in materials and statistics/data science and professional skill modules before the start of the year.

In the first year of study, students must take foundational courses in materials and data science, ensuring that all students have fundamental training outside their home discipline, and a new survey course with hands-on modules demonstrating applications of AI in materials science.

In the second year, aiM Fellows will work in cross-disciplinary pairs in a capstone “AI in Materials” project course on customized projects based on real-world problems that may be related to their dissertation topics. These projects will provide invaluable experience in not only applying their growing technical expertise but also in practicing interdisciplinary team science, a hallmark of their careers.

Following the second-year project, students will do a three- or six-month internship at a national laboratory or company partner, gaining exposure to world-class facilities and experience in approaching real-world problems with data-driven materials science.

Along the way, aiM Fellows will receive professional development opportunities, such as mentoring, workshops, and experiences to develop their leadership and communication skills.