How AI Helps to Advance New Materials Discovery

In the field of new materials, there has recently been an exponential growth in the generation of theoretical and experimental materials data. The extraction of actionable knowledge from this rich data, however, has emerged as a grand challenge that invites not only the application but also advancement of the state-of-the-art in computational science. There are barriers to this integration—namely, the need for domain-specific knowledge and the incorporation of heterogeneous data. These challenges offer opportunities to make foundational advancements in both fields.

R. Bruce van Dover, Materials Science and Engineering (MSE), is leading a team of Cornell researchers from MSE and Computer Science, with collaborators at the California Institute of Technology, University of Colorado, and Northwestern University, to integrate materials science and computer science methods to dramatically accelerate, by orders of magnitude, the discovery and development of new materials. The central innovation is a multi-agent system called the Scientific Autonomous Reasoning Agent (SARA).

SARA will be able to help researchers identify promising new candidate materials, guide their synthesis, and establish their value for solving prominent materials challenges. The system will integrate, in a synergistic and complementary way, first principles quantum physics, experimental materials synthesis, processing, and characterization, and AI-based algorithms for reasoning and conducting science, including the representation, planning, optimization, and learning of materials knowledge. It will be augmented significantly with human insights so that the AI leverages that of expert scientists, creating an unprecedented platform for human-machine collaboration. SARA will usher in a new era of scientific discovery, while also demonstrating the vast benefits of cross-disciplinary collaboration for both disciplines.

Cornell Researchers

Funding Received

$7.5 Million spanning 5 years

Sponsored by

Other Research Sponsored by United States Department of Defense, Air Force Office of Scientific Research