Google DeepMind’s artificial intelligence (AI) has achieved a significant milestone by predicting the structures of more than 2 million novel chemical materials, presenting potential advancements in real-world technologies. The findings were detailed in a paper published in the Nature Journal on November 29.

The AI, developed by DeepMind, utilized data from the Materials Project, an international research consortium founded in 2011 at the Lawrence Berkeley National Laboratory. The dataset incorporated information on around 50,000 existing materials.

The paper highlighted the cost and time challenges associated with identifying and creating new materials, citing the two-decade timeline for lithium-ion batteries to become commercially viable. The research scientist at DeepMind, Ekin Dogus Cubuk, expressed optimism about leveraging advancements in experimentation, autonomous synthesis, and machine learning models to significantly shorten the discovery and synthesis timeline, which traditionally takes 10 to 20 years.

The AI’s training involved predicting the stability of these novel materials, with nearly 400,000 designs potentially heading for laboratory testing. The envisioned applications of this research span the development of batteries, solar panels, and computer chips with improved performance.

DeepMind has expressed its commitment to sharing its data with the research community to accelerate progress in material discovery. However, Kristin Persson, director of the Materials Project, noted the industry’s cautious approach due to potential cost increases, emphasizing that new materials often take time to become cost-effective. Shrinking this timeline would represent a significant breakthrough, Persson added.

Moving forward, DeepMind aims to focus on predicting the synthesizability of these novel materials in laboratory conditions. The integration of AI in material discovery holds promise for revolutionizing technology and potentially expediting the development of cutting-edge products.