The Electric Flash Analysis: AI Can Make Better EV Batteries, but Only If the Industry Fixes Its Data Problem

A reminder to tune in tomorrow at 11:30 a.m. PT to a special pop-up event: Will AI Transform Batteries? Machine learning has unleashed a new age for drug discovery and on-line advertising, but it has barely penetrated battery science. Can it do so this decade? My guests for this vibrant Live Chat are AI pioneer Andrew Ng, founder of Landing AI, and Tim Holme, CTO of QuantumScape. RSVP here and pose a question of Andrew and Tim yourself. 

Machine learning software that finds patterns in vast stores of data has transformed drug discovery, online advertising and social media, but not so much battery science. The discoveries involving batteries for electric vehicles have come almost entirely from iterative, hunt-and-peck methods.

Two new research papers suggest that a lack of organized data, the fodder of successful machine learning, is at fault in the battery field's slowness to adopt artificial intelligence.  

Since the birth of the lithium-ion age four decades ago, researchers have largely failed to methodically store and label the results of their tens of thousands of experiments—the dead ends, fiascos and successes. Even when they have compiled their data, hyper-secretive companies, government labs and university professors have been prone to hold on to even the most innocuous results, fearful that someone would somehow steal a march on them.

In the two new papers, battery scientists from prominent labs are beseeching colleagues to break the data drought and be more methodical with the quality of research they produce. The potential payoff of the new approach will be the subject of The Electric’s next live chat, Will AI Transform Batteries? It is scheduled for Wednesday at 11:30 a.m. Pacific Time. I will host artificial intelligence pioneer Andrew Ng, whose startup Landing AI develops machine vision software for quality control in manufacturing, and Tim Holme, chief technology officer of QuantumScape, a $10 billion lithium-metal battery startup.