Help your computer build the next computer
Google launched AutoML, the first machine learning program which learns to perform particular tasks alone and analyzes data autonomously. In short, AutoML is building the first independent AI to write it’s first machine learning software, more sophisticated and powerful than human systems.
We live in an era of wonders. Yet tech wonders involve a lot of resources, with the most valuable and scarce being the human researchers. One might think that the AI field is crawling with bespectacled geniuses, writing the code of the century, when in fact “good” AI experts are in high demand. Leading industry giants such as Google and co. are fighting a ruthless fight to monopolize intelligence teams, spending and investing with millions of dollars on top candidates. Yet even finding the right person is not a guarantee of success which is why most AI researchers spend a number of years in boot camp, dreaming of programming complex machines, without getting even close to the real thing.
Dark Arts 2.0
But what makes their jobs so hard you ask? Well, Neural Network programming can be a….drag. It’s surely not like these wiz guys are simply building a website or debugging a software. We are talking about running hundreds of experiments across a vast network of machines and poking and prodding it on each end, to see how well it can recognize an image or translate from one language from another. And this is only step one of trial and error, followed by significant time and resources invested in doing the same thing over and over until they create a fail-safe.
In short, we are talking about a completely new kind of looking at computer programming, with experts needed to have extensive knowledge, math skills, intuition and vision. Researchers joke among themselves, saying they’d be practitioners of a “dark art”.
AutoML
Change is a curious thing. Google’s top engineers took the podium to talk about the changes their company was going through and highlighted their most recent project: AutoML, a machine learning program. But unlike other machine learning systems, ML is the machine to rule them all. Well, not quite but it can build its own machine-learning algorithms. Google is hoping to lead the revolution and create future AI overlords, by taking the buggy human element out of the algorithm.
ML can as of now classify tons of images and recognize and name objects. It can design its own small neural networks that are similar to the ones built by humans.
The aim is to accelerate the progress of the AI field for online and offline industries. With the programs diversifying and rapidly expanding, new markets are possibly soon going to open up. At Beluga, we are very keen to know what the future holds. And in a couple of years, looking back, these spearheading discoveries will look like sandbox games.
About Beluga
Beluga helps fast-moving companies to translate their digital contents. With more than a decade of experience, professional linguists in all major markets and the latest translation technology at use, Beluga is a stable partner of many of the most thriving enterprises in the technology sector. The business goal: To help fast-growing companies offer their international audiences an excellent and engaging user experience.
