Main image of article Google Wants to Automate A.I and Machine Learning
Companies are obsessed with the potential of artificial intelligence (A.I.). Even those executives who understand precious little about it want their divisions to utilize it. And therein lies the rub: despite the enormous hype around A.I., there are precious few tech professionals who possess the technical knowledge necessary to implement, maintain, and iterate on an A.I. platform. That’s great for A.I. specialists, who have their pick of jobs (and incredible salaries and benefits), but it’s not so wonderful for companies that can’t find the necessary talent at the right price. “And if you’re one of the companies that has access to ML/A.I. engineers, you still have to manage the time-intensive and complicated process of building your own custom ML model,” Fei-Fei Li and Jia Li, chief scientist and head of R&D, respectively, of Google’s Cloud AI division, wrote in a lengthy Jan. 17 blog post. “While Google has offered pre-trained machine learning models via APIs that perform specific tasks, there's still a long road ahead if we want to bring A.I. to everyone.” In that spirit, Google is rolling out Cloud AutoML, which it bills as a way to automate the creation of customized machine-learning models. The first product on the new platform, Cloud AutoML Vision, will offer companies and developers a drag-and-drop interface for creating image-recognition models. In their blog post, Fei-Fei Li and Jia Li suggested that Cloud AutoML produces more accurate results than generic machine-learning APIs; but considering how the platform is still in Alpha, it could take some time for Google to iron out any issues. It remains to be seen whether any platform can effectively automate something as complicated, nuanced, and ultimately mission-critical as A.I. and machine learning. If Cloud AutoML succeeds, then other companies may rush to produce tools that reduce the creation of A.I. models to user-friendly dragging-and-dropping. However, even if A.I. automation goes mainstream, most firms will almost certainly prefer to hire at least one dedicated specialist to oversee the algorithms.