Electric-Car Batteries Get a Boost From Artificial Intelligence

At Carnegie Mellon University, machine-learning software enabled a robot named Otto to discover a new water-based electrolyte solution that performs at a higher voltage than the current standard. Otto made a new mixture of water and four salts every seven minutes, tested the mixture for voltage stability and sent the results to the cloud.

In 40 hours, Otto, using AI, found a recipe for an electrolyte “that was better than state-of-the-art,” says Venkat Viswanathan, an associate professor of mechanical engineering at Carnegie Mellon and a co-author of the study, which is being published this month in Cell Reports Physical Science. To find the same recipe using standard methodology, he says, would have taken years.