According to the data released by DeepMind, AlphaGo has 12 layers of neural networks, in which millions of humanoid neurons are connected with one another. Moreover, AlphaGo has a killer feature: deep learning. Simply speaking, we can set a goal for artificial intelligence so that it will search for the best solution among a massive amount of data. In the past, we had to make the computer accurately execute instructions made by human beings through programming; however, the breakthrough of deep learning lies in that the searching task has been transferred to neural networks and it is artificial intelligence that tries to find among numerous data the best solution, which may have never occurred to people before.

AlphaGo has possessed the ability to defeat South Korean Go player Lee Sedol through a deep learning of numerous Go manuals and through competition against itself again and again. However, once people make a move which is a “blind spot” for its data, it would perform in the same way as in the fourth game. The worst Go player can be just one step away from the best Go player, and the natural moat between them is the big data. The control of big data may be a major way for human beings to control intelligent robots.

Actually, artificial intelligence represented by AlphaGo is still in its infancy. It still has a long way to go before it can completely replace human beings, for there are still many technical problems to solve. Even if artificial intelligence develops by leaps and bounds and may have a substitution effect for the present labor force in decades, it would not lead to a high unemployment rate or a large-scale social instability. Throughout the history, every important technical progress involves as great a development in civilization and technological innovation as the breakthrough of artificial intelligence. Although it indeed has a substitution effect for labor force to a certain extent, new technical development greatly expands the industrial dimension so that many new subdivided industries derive from it and many more people get employed.

While probing into the future of artificial intelligence, many people cannot help worrying about the destiny of human beings, asking whether human beings would be enslaved by computers if artificial intelligence comes to an advanced phase of development. Is this an urgent question of human beings’ ultimate destiny or just an unnecessary concern? There is still no standard answer. What we know is that artificial intelligence is still limited to the framework of “machine learning”, and human beings still have a strong control over artificial intelligence through their control of the big data. The concern mainly comes from people’s ignorance of newly emerging things. Instead of sleeplessness, we should try to find the boundary of artificial intelligence as soon as possible and form a common understanding and control regulation. As for China, we should remain calm, take advantage of the development of artificial intelligence and boost the technological innovation with Chinese characteristics.

Note:
The author is an Associate Professor in Tongji, SEM. This article originally appeared in Jiefang Daily on the following website: https://newspaper.jfdaily.com/jfrb/html/2016-03/22/content_181300.htm