Deep learning + big data "Artificial intelligence" standard needs to b

AI is a hot word now. According to the survey results of Bloomberg information research company, in the past two years, the frequency of AI in the conference call records of many companies has increased dramatically, and the intention to acquire AI has also increased. However, people's understanding of artificial intelligence is not the same. Any software, machine and so on that are connected with automation are often put into the big basket of artificial intelligence.

On June 7, 2017 local time, at the "artificial intelligence" global summit held by the International Telecommunication Union (ITU), Sophia, a humanoid robot developed by Hansen robot company, appeared in Geneva, Switzerland.

At present, people's understanding of artificial intelligence mostly comes from the artistic images in novels and movies, from the cold-blooded imitation of real person David in the movie alien to the cute robot C-3PO in Star Wars. These images often have the same shape, thinking mode and even feelings as human beings, and they are more powerful than human beings. Based on this understanding, AI is often equated with robot; "Alpha dog" makes some people draw an equal sign between AI and computer program; even before twitter, simple optimization of content is also considered by some people that Twitter is trying to "make AI more intelligent"

To discuss the standards of artificial intelligence, we have to start with the birth of computers. Since the 1950s, "Turing test" has been generally regarded as the standard to judge whether a machine can be called artificial intelligence. But for a long time, artificial intelligence only stays in the research stage. The third artificial intelligence upsurge started in 2006 has pushed artificial intelligence from the laboratory to practical application and created real value. Artificial intelligence has become a new investment outlet.

The core of artificial intelligence lies in the convergence of three technologies: first, the evolution of algorithm with deep learning as the core; second, the improvement of computer processing ability and computing speed, which can effectively identify characters, listen to sounds and distinguish pictures; third, the development of Internet technology, which produces a wide range of massive data. With its excellent computing power, deep learning method and massive data as its material, the "machine" has significantly improved its intelligence. As Charles Isbell, a researcher at Georgia Institute of technology, described, true artificial intelligence requires computer programs or machines to demonstrate self-management, emotional expression and innovation. Therefore, the current AI = deep learning + big data, which is the most epochal and widely accepted understanding.

Artificial intelligence does not always appear in the form of robot, in many cases, it even exists in the form of invisible. The voice assistant "Siri" of Apple mobile phone just hears its voice and doesn't see its shape. On the contrary, there are many machines in the form of robots or even humanoid robots that do not reach the level of artificial intelligence. Moreover, artificial intelligence is not a super human brain, because there is a huge difference between its working principle and that of human brain.

The purpose of defining AI is to make AI develop more orderly. We can neither deify artificial intelligence, let the industry lose the minimum access threshold, enterprises can casually rub "hot spots", and stick "artificial intelligence" label for their various products at will; on the other hand, we don't need to demonize it, at least in the visible future, artificial intelligence can't be superior to human beings, let alone have the terrible consequences of destroying human beings. In the discussion of the nature of artificial intelligence, Ian bogost, a professor of interactive computing at Georgia Institute of technology, may be closer to the truth: they are devices made by human beings, running software written by human beings, full of their advantages and disadvantages. Of course, such a view does not change the fact that artificial intelligence becomes the next technological frontier and industrial outlet, nor weaken its potential to break the world order.