Los Alamos National Laboratory is installing the latest OpenAI o-series models to its Venado supercomputer to further upgrade its artificial intelligence capability for national security research. The partnership with OpenAI in the installation is hoped to yield expert reasoning on a wide range of complex scientific studies, the lab said Thursday.
The Venado computer will operate on a classified network for secure sharing with researchers from the Los Alamos, Lawrence Livermore and Sandia national labs of the Department of Energy.
According to Thom Mason, the laboratory’s director, national security threats have grown more complex and more urgent, requiring advanced technological approaches.
“Artificial intelligence models from OpenAI will allow us to do this more successfully, while also advancing our scientific missions to solve some of the nation’s most important challenges,” the lab official stressed.
Wide-Ranging Tech Collaboration
Los Alamos already had previous collaboration with OpenAI on projects to improve AI safety. In July, the lab and OpenAI launched a study to evaluate the risks of advanced AI being used to bring about biological threats.
The Energy Department and its national laboratories are working to strengthen collaboration with other technology companies for various applications, including new materials design and quantum algorithm development.
For the Venado supercomputer in Los Alamos, NVIDIA provided its GH200 Grace Hopper Superchips, designed to process high-performance computing and large-scale AI applications. In November 2023, the Energy Department’s National Nuclear Security Administration and the Los Alamos lab also tapped SambaNova Systems in a multi-year partnership for its data analytics and artificial intelligence software suite to expand the lab’s large language models and general AI technologies.