Google has just announced that its artificial intelligence (AI) project AutoML, which was designed to help create other AIs, has outperformed human AI engineers by building machine-learning software that is more efficient and powerful that human-designed systems. Recently, an AutoML system scored 82 percent for categorising images by content, setting a new record. Although a relatively simple task, AutoML also beat a human-built system at marking the location of multiple objects in an image — a more complex task that is integral to autonomous robots and augmented reality, scoring scored 43 percent versus the human-built system's 39 percent. This result is highly significant, as few people posses the expertise to build next generation AI systems, and once this area has been automated, it could have profound implications for the industry. To find out more check out Google's learning software learns to write learning software.