In recent years, especially in March 2016, alphago developed by Google’s deepmind company defeated Li Shishi in the ninth stage of Korean professional go, making artificial intelligence (AI) technology hot and sweeping the world. “Artificial intelligence” has become a fashionable term, which is favored and sought after by enterprises and capital. Smart manufacturing, smart oil drilling, smart travel, smart city, smart water and other concepts also emerge one after another. Artificial intelligence has indeed brought great convenience to human production and life.
However, then, doubts about artificial intelligence came one after another. The ethical problems brought by artificial intelligence, the violation of human privacy, and whether it will seize human employment opportunities have begun to be paid more and more attention by the public.
What is artificial intelligence
Literally, artificial intelligence is an intelligent machine made by human beings. In fact, there are different definitions of artificial intelligence. Dr. John McCarthy, the winner of the Turing Award for artificial intelligence, once defined artificial intelligence as “the science and engineering of manufacturing intelligent machines”. Ma wennski, the father of artificial intelligence and the founder of framework theory, defines “if people do things that require intelligence and the ability to solve problems.”
In fact, artificial intelligence technology integrates the knowledge of information theory, cybernetics, automation, bionics, biology, psychology, mathematical logic, linguistics, medicine and philosophy, and is presented by computers and other machines.
History of artificial intelligence
In 1946, the first computer was invented, which created the history of Turing machine, that is, the British mathematician Alan Turing proposed to generate logic signals (0 and 1) by manipulating circuit switches. People can create machines comparable to perfect mathematicians. The computers we currently use are developed based on this theory. Since then, people began to believe that computers can develop into intelligent machines to replace human beings.
By the 1970s, because technicians underestimated the technical difficulty of artificial intelligence development and people’s panic about artificial intelligence, countries began to tighten the development of artificial intelligence, and artificial intelligence fell into a trough.
Subsequently, the application of some mathematical algorithms has made a certain development of artificial intelligence. However, the emergence of Internet of things and search engine has a great impact on artificial intelligence. People used to think that the expert system information can be obtained through the Internet. The rise of artificial intelligence again benefits from the development of algorithms such as deep learning and GPU hardware technology. In short, thought and science and technology complement each other, so as to continuously promote technological progress and create the future.
Mathematician Alan Turing invented a method to test whether a machine is intelligent. In short, it is to test the objects in the blind box. If 30% of people can’t distinguish whether it is a machine or a person in the blind box, the machine passes the test. This test method was put forward shortly after the emergence of artificial intelligence and even computers at that time. The rationality of this method remains to be discussed. Even from a philosophical point of view, what does it mean if it’s human and doesn’t pass the test?
As the definition and test methods of artificial intelligence are controversial, it makes the current artificial intelligence become a gimmick of many businesses and capital.
Composition of artificial intelligence
The three elements of artificial intelligence: algorithm, computing power and data.
The algorithm itself is a superposition of multi-disciplinary mathematics such as biology and logic. The history of the algorithm itself is earlier than the invention of artificial intelligence and computer. In fact, in order to adapt to the low computing power of early computers, the early algorithms were studied very complex, so as to reduce the load of computers by optimizing algorithms.
According to Moore’s law, “the number of transistors that can be accommodated in an integrated circuit will double every 18 months “. in other words, the performance of the processor doubles every two years. The iterative update speed of the computer hardware level is accelerated, resulting in the continuous improvement of computing power. The development of neural network algorithm is due to the improvement of GPU computing power. It is found that GPU is more suitable for this flat operation than cup in neural network algorithm operation.
The development of sensor, Internet of things and mobile Internet technology provides massive data for artificial intelligence data analysis, which also speeds up the pace of artificial intelligence.
In a sense, due to the improvement of computing power, the development of algorithms in recent years is not as deep and complex as in the case of low computing power in the last century. The reason is very simple. The general algorithm model can be compensated by strong computing power, at most adding hardware or time.
Artificial intelligence classification
In terms of the ability level of artificial intelligence, it can be roughly divided into weak artificial intelligence, strong artificial intelligence and super artificial intelligence.
Weak artificial intelligence is our most widely used machine learning model, which mainly focuses on the content of a certain field, such as go, predicting driving time, etc.
Strong artificial intelligence focuses more on robots with the same intelligence level as people, which is not only algorithms, but also human function, automatic control and so on.
Super artificial intelligence, like the matrix, surpasses the human level and even controls the human level.
From the current development situation and the principle of artificial intelligence, human beings are digging deep into weak artificial intelligence, and there is a trend towards strong artificial intelligence, but the possibility of super artificial intelligence is almost zero – artificial intelligence is not life, and there is no silicon-based life.
Artificial intelligence and machine learning
Expert system, machine learning, evolutionary computing, fuzzy logic, computer vision, natural language processing and recommendation system all belong to the category of artificial intelligence. However, at present, machine learning is the most widely used.
The most used concepts in machine learning: deep learning (neural network), reinforcement learning, supervised learning, unsupervised learning and so on.
According to the scientific way of thinking, the development or implementation of anything can be divided according to itto (input, tool / technology, output).
Conventional computer mode: data + rule = answer;
The way of machine learning: data + answer = specification
Deep learning, that is, neural network, achieves the effect of image recognition or sound recognition by simulating the working principle of human neural network (which is actually very different from that of human neural network).
Reinforcement learning, in the case of not having a lot of data but knowing the rules, machines and themselves carry out continuous training to improve their level. The simplest example is alphago, which is reinforcement learning.
Supervised learning is to give reward feedback and confirm the learning effect in the learning process.
Development trend of artificial intelligence
In the final analysis, artificial intelligence is a tool made by human beings to help human beings improve efficiency in production and life. We can simply imagine artificial intelligence as an invention similar to television and refrigerator. From the perspective of human history, the emergence of a new technology will not seize human employment, but will create more employment opportunities.
From the perspective of God, the whole universe is a complex system created by the creator. It is impossible for machines made by human beings to surpass human beings themselves with some algorithms. Only a neuron in human beings needs several layers of complex operations of neural network algorithms for simulation. Therefore, artificial intelligence can only be used as a tool to benefit human beings, which of course needs to be regulated by laws and regulations Without the ability to create super artificial intelligence to control human beings.
However, when human beings rely too much on artificial intelligence, in fact, it gradually destroys human innovation, exploration and even survival ability. Imagine how many people will get lost if there is no navigation? How inconvenient it will be for lunch if there is no mobile phone to order. From this perspective, it is not that artificial intelligence is becoming smarter and smarter, but that human beings are becoming more and more degraded. How to use technology In the wave of development, let people keep real life and interesting soul?