Google DeepMind’s artificial intelligence (AI) has achieved a milestone by predicting the structures of more than two million novel chemical materials.
This development, disclosed in a recent Nature paper, signifies a leap in advancing real-world technologies.
In their scientific publication on Wednesday, 29 November, the AI company reported that nearly 400,000 of its theoretical material designs are poised for imminent laboratory testing.
Potential applications of this research span across the enhancement of batteries, solar panels, and computer chips to improve overall performance.
The paper underscores the traditionally costly and time-intensive nature of identifying and creating new materials.
It took approximately two decades of research before lithium-ion batteries, now ubiquitous in devices like phones, laptops, and electric vehicles, became commercially viable.
Ekin Dogus Cubuk, a research scientist at DeepMind, expressed optimism:
“We’re hoping that big improvements in experimentation, autonomous synthesis, and machine learning models will significantly shorten that 10 to 20-year timeline to something that’s much more manageable,”
The AI developed by DeepMind underwent training using data from the Materials Project, an international research consortium established at the Lawrence Berkeley National Laboratory in 2011.
The dataset comprised information on around 50,000 preexisting materials.
Commitment to Research Community
DeepMind has committed to sharing its data with the research community, aiming to accelerate further breakthroughs in material discovery.
However, Kristin Persson, director of the Materials Project, mentioned in the paper that the industry remains cautious about potential cost increases, highlighting that new materials often take time to become cost-effective
“Industry tends to be a little risk-averse when it comes to cost increases, and new materials typically take a bit of time before they become cost-effective. If we can shrink that even a bit more, it would be considered a real breakthrough.”
Following the AI’s success in predicting the stability of these novel materials, DeepMind has now shifted its focus to forecasting their synthesisability in laboratory conditions.