The University of Michigan, together with Los Alamos National Laboratory and Argonne National Laboratory, has launched Mission Lodestar, a national initiative to demonstrate AI’s potential for scientific discovery, supported by OpenAI’s contribution to the U.S. Department of Energy’s Genesis Mission. Its first focused test case is Project Huron, led by our director Venkat Viswanathan, which uses AI and advanced computing to discover hydride superconductors that reach high temperatures at moderate pressure.
Named for the Huron River, the project begins as a focused 18-month effort centered on a small team in Ann Arbor and a network of collaborators at national laboratories and universities. At its core is an AI scientist: a team of AI agents that plans a discovery campaign, runs simulations from fast machine-learning models up to first-principles calculations, and learns from every result, with human scientists in the loop.
“Recently, there have been discoveries of two new hydride materials, which is extremely rare in materials discovery,” said Viswanathan. “These materials were first computationally predicted and subsequently experimentally verified. What is exciting is that those predictions were made using computational tools far less sophisticated than the AI available today. This leap in AI capability presents an enormous opportunity.”
Read the full announcement here.
