Listen : Audio version of this article
AI is literally breaking grounds and Google has gone a step further in this process. As all industries are trying to incorporate AI in their processing, marketing and many more domains. Since skilled labour regarding AI is yet to arrive, AI itself can solve the issue as machine learning takes over skilled labour to automate the process of creating machine learning software. In a recent announcement, Google announced that they have developed something called AutoML, a brainchild of the AI research group of Google Brain.
As the company professes, the trickiest bits of machine learning software design can now be designed by AutoML that will perform a specific task. So much so, even the best human experts will be beaten for skill when it comes to machine learning. In short, machine learning has just excelled itself. Revolutionizing what developers can do and their limits, AI can now design itself and that too better than its creators.
Investing skills elsewhere
Now that AutoML has entered the scenario, less expertise will be needed to develop particular software for a specific task. Hence, developers can invest more time in bettering machine learning. Cloud computing can now easily host machine learning and build great software in unison. All of these are efforts to draw new customers as Amazon and Microsoft is racing ahead in the AI business and this may well change the game.
Deep learning is at the heart of AutoML since Google is focusing more and more on speech recognition, image processing, robotics as well as translation. It is going to make smarter software using ANN or artificial neural network. However, the architecture of an ANN is particularly tricky. Intuitional decisions have been the process as of now, which may change now. A machine learning system apparently can do it better than humans when it comes to languages and matches the best architectures regarding images. Surely, that is unprecedented power in your hands.
AI can do more than replicate
While AI was initially made by Google to do work for developers, it can apparently do it better and most importantly, innovate if need be. During the image and language processing architecture building experiment, it was found that AI had put forward a new process of building architecture that was never thought of regarding these processes. AI is no longer learning, it is learning and converting knowledge into innovation. However, Google is not alone in this venture. OpenAI, an NPO supported by Elon Musk and people at academia are doing similar things.
However, that would never mean that the technique is going to become available immediately. The amount of power that has been consumed during the processes and the amount of investment that has been done can only be possible because of Google. However, a cost-effective method shall be devised in good time and a larger team is at work to make sure that it requires less resources and it can further function without much human instruction. In short, the desired goal is achieved, but its spread will take time.