As a kind of science and technology, artificial intelligence must follow the laws of natural science in its development process. Depending on its actual conditions and development mechanism, artificial intelligence can be recognized and predicted. At present, the specific technical routes of artificial intelligence are diverse, and the future development is full of possibilities, but its development trend still has "traces" to be found.
From a broad perspective, there are three main trends:
The first is to have high performance. The new round of artificial intelligence based on big data and deep learning has largely benefited from the increase in computing power. Without a huge increase in the computing power of supercomputers, it is impossible to complete the task of processing massive data. At present, the world's number one supercomputer in computing power is the “Summit” of Oak Ridge National Laboratory in the United States, and the floating-point computing speed can reach 2 billion times per second. However, with the constraints of chips, energy consumption and other factors, it is becoming more and more difficult to increase the computing power of supercomputers. People have begun to look for other alternatives, and quantum computing has gradually entered people's field of vision. In May 2017, Chinese scientists developed the world's first optical quantum computer that surpassed the early classic computers and achieved 10 superconducting qubit entanglements. In August 2019, scientific research teams such as Zhejiang University and Institute of Physics of the Chinese Academy of Sciences jointly developed Superconducting qubit chip. On October 24 this year, Google announced that they used a 54-bit qubit quantum computer to complete the world's fastest supercomputing task that requires 10,000 years of computing within 200 seconds. This initially shows the subversive level of computing power in quantum computing, and the "quantum superiority" of concern from all walks of life has been achieved. Although this achievement is still confined to a specific field and there is still a long way to go before it is practical and general, its significance is significant, meaning that the computing power foundation of artificial intelligence can undergo fundamental changes, and artificial intelligence will undoubtedly present high performance.
Second, it is universal. The relevant capabilities of the current artificial intelligence in specific fields, such as complex computing, image recognition, and speech processing, have far surpassed humans. But its limitations are also obvious, that is, they cannot be universally used, and artificial intelligence in one field will become "artificial retardation" when it is in another field. To solve the general problem of artificial intelligence, we must develop strong artificial intelligence so that machines can truly think like humans. The Turing Award winner Judia Poel elaborated on causality in the book Why, and divided it into three levels: "association", "intervention" and "counterfactual reasoning." Machine learning is only at the lowest first level, the weak artificial intelligence stage. There are many methods for machine learning, and deep learning is just one of them. The third level of "counterfactual reasoning" is the product of human imagination and is a unique ability of human beings, namely, strong artificial intelligence. Pearl's causal theory opened a window for the research of strong artificial intelligence, and theoretically pointed out the general stage and the direction of the development of artificial intelligence. It is very likely to open up a new realm of algorithm theory innovation and development.
The third is to have reliability. In the future, artificial intelligence must have good interpretability, so that its learning models and corresponding decisions can be understood by human users, thereby increasing people's trust in artificial intelligence systems. However, the current machine learning technology has not been able to meet people's expectations and requirements. Even if the machine has already reached a conclusion, users often cannot help but make a question mark in their hearts and feel that they must be rechecked manually. Especially when these artificial intelligence technologies are applied to assist decision-making, people's incomplete trust will multiply even more. The needs of users are the driving force and inevitable direction of the development of artificial intelligence technology. It is foreseeable that in the near future, the cross-fusion between artificial intelligence technologies will be deepened, and cross-fusion between different disciplines will be more frequent. By integrating the advantages of various aspects and disciplines by taking advantage of each other's strengths and weaknesses, it is expected to meet people's requirements for the reliability of artificial intelligence systems.
Contact: Manager Xu
Phone: 13907330718
Tel: 0731-22222718
Email: hniatcom@163.com
Add: Room 603, 6th Floor, Shifting Room, No. 2, Orbit Zhigu, No. 79 Liancheng Road, Shifeng District, Zhuzhou City, Hunan Province