The world of technology is set for a seismic shift, courtesy of Google DeepMind's artificial intelligence (AI) tool, GNoME.
This nifty little wizard has identified a staggering 2.2 million new crystal structures, potentially paving the way for advancements in everything from renewable energy to advanced computation.
This treasure trove of theoretically stable but experimentally unrealised combinations is a whopping 45 times larger than the number of such substances discovered in the history of science. The researchers are now planning to make 381,000 of the most promising structures available for fellow scientists to test their viability in various fields.
The beauty of this venture is that it shows how AI can cut through years of experimental work, potentially delivering improved products and processes in a fraction of the time. As Ekin Dogus Cubuk, a co-author of the paper, puts it, "Materials science is where abstract thought meets the physical universe."
The DeepMind team used machine learning to generate candidate structures and then gauge their likely stability. The number of substances found is equivalent to almost 800 years of previous experimentally acquired knowledge. This significant leap forward could revolutionise industries from microchips to photovoltaics, breaking the bottleneck caused by expensive trial-and-error approaches.
The potential applications of these new compounds are vast, including the invention of versatile layered materials and the development of neuromorphic computing, which uses chips to mirror the workings of the human brain.
Researchers from the University of California, Berkeley and the Lawrence Berkeley National Laboratory have already started using these findings to create new materials. Their autonomous laboratory, known as the A-lab, has managed to create 41 novel compounds from a target list of 58, boasting a success rate of over 70%.
This breakthrough in the world of technology is a testament to the power of AI and its potential to revolutionise our world. The future is here, and it's crystal clear!
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai
Comentarios