According to a recent article in the New York Times, only about 10,000 people in the world have the know-how to build the complicated algorithms needed to create artificial intelligence. But those numbers can't keep up with potential demand. Because of the too-small pool of qualified experts, some companies are developing tools that essentially allow AI to build AI.
Like Google's new AutoML technology. According to the Times, “Google said AutoML could now build algorithms that, in some cases, identified objects in photos more accurately than services built solely by human experts.” Experts call this trend "learning to learn," or meta-learning, and these experts believe it will someday work for things like speech recognition or machine translation.