The Era of Deep Learning

Written by: Gal Kalinsky

Deep learning is part of a broader family of machine learning methods. It includes algorithms that are inspired by the way the human brain works. It uses artificial neural networks for learning, sorting and processing data, in the same manner as the organic brain does. The bigger the data set it processes and the more computation power it consumes, the more accurate its analysis gets. Deep learning technologies are applied to many fields, such as computer vision, understanding of natural languages and even speech recognition.

In 2011, during the “German Traffic Sign Recognition Benchmark”, a German competition, a deep learning algorithm obtained results that were twice as good as would be expected from actual people. Since then, a lot has happened. Google started using deep learning for many of its services, one of the services is Natural Language Processing. Google Assistant speech recognition AI uses deep neural networks to learn how to parse and understand verbal commands and respond to questions accordingly.

Not to be outdone, in October 2017, Microsoft Chief Research Officer Rick Rashid demonstrated an online translation service: As he was talking, the service recognized his English sentences, translated them to Mandarin and synthesized them with his own voice. The translation was done almost instantly with an error rate of only 7 percent.

Although the opportunities created by the use of deep learning technologies are very exciting, we must be concerned about the consequences that may arise. Deep learning algorithms learn from the data that is provided. Given that it is often difficult or even impossible to understand the process that the deep learning algorithm uses to apply the data, it is important that the data we will be correct at the outset. Failing to provide valid information can affect human lives:

Further, society needs to decide how much control we can knowingly cede to a computer that has no morals and no ethics. For example, Microsoft’s newly launched AI powered Twitterbot Tay rapidly devolved into an offensive and racist tweeter.
In addition, services that are at risk of being replaced by AI, like cashiers, security guards, analysts, cab drivers, or even doctors, are services that provide employment for millions of people all over the world.

There is no doubt that computers can perform many tasks better than humans. Now is the right time to start thinking about how we are going to adjust our social and economic structures to be aware of this new reality and maximize its potential as safely and conscientiously as possible.