Nissan’s Silicon Valley lab is using AI and machine learning to accelerate the development of new materials for future vehicle applications. This is done by simulating the properties of materials for a lot more cases than can be tested experimentally, in a short time. This can help sample millions of materials, and then screen for candidates based on the properties that are desired. AI and machine learning can drastically reduce the time needed for this process, making it about 1,000 times faster than traditional research methods.
Nissan has identified select internal projects to observe and evaluate applications of AI, and the work being done at NATC-SV is part of the company’s larger effort to quickly and safely bring the latest tech innovations to drivers. The AI and machine-learning applications being implemented at NATCSV are distinct from generative AI tools, such as ChatGPT. The key difference is that generative AI creates new material, while machine learning uses existing material to make predictions.
Experimentation remains key, and the Materials Informatics team sets benchmarks for the properties of a new type of material and runs AI-optimized simulations to test those factors. While researchers still experimentally validate all results, it makes the work more efficient.
The team is also adding more scientific equipment to its lab to be able to conduct additional research and testing on physical material samples. Machine learning and AI are good as screening tools, but researchers still use some of the classical, physics-based approaches to have final confirmation of the materials’ properties.
Leveraging AI and machine learning has not eliminated the need for experienced, skilled researchers, who use AI as a tool but rely on their own knowledge to make final decisions. Interpreting the simulations is still done by real humans.
As a result, the Materials Informatics team is continuing to grow and is hoping to further expand its staff to keep up with the tremendous amount of data and analysis needed.