Google DeepMind leveraged artificial intelligence (AI) to forecast the structure of over 2 million new materials, a groundbreaking achievement with potential applications to enhance real-world technologies, according to a research paper published in the science journal Nature on Wednesday. The AI firm, owned by Alphabet, anticipates that nearly 400,000 of its conceptualized material designs could be produced in laboratory conditions in the near future. This advancement holds promise for the development of improved batteries, solar panels, and computer chips.
The conventional process of discovering and synthesizing new materials is both time-consuming and costly. The breakthrough could potentially accelerate the timeline for bringing new materials to market. Ekin Dogus Cubuk, a research scientist at DeepMind, expressed optimism about significant reductions in the 10 to 20-year timeline through advancements in experimentation, autonomous synthesis, and machine learning models.
DeepMind’s AI was trained using data from the Materials Project, an international research group founded in 2011 at the Lawrence Berkeley National Laboratory. The group compiled data from approximately 50,000 known materials. DeepMind intends to share its data with the research community, aiming to expedite further breakthroughs in material discovery.
Kristin Persson, director of the Materials Project, highlighted the potential impact on industry, noting that if the time and cost associated with developing new materials can be further reduced, it would be considered a significant breakthrough. DeepMind plans to shift its focus from predicting the stability of new materials to forecasting their ease of synthesis in laboratory settings.
