Xu Li: Machine guessing or smashing Newton's apples, Shanghai's tolerance and openness are the best soil

The Surging News reporter Zhang Jing Yang Yang

On July 8, the 2021 World Artificial Intelligence Conference opened in Shanghai. Xu Li, co-founder and CEO of SenseTime, said at the opening ceremony that today's artificial intelligence conjecture and machine conjecture may be the apple that Newton smashed on the head, and Shanghai's inclusiveness and openness is its best soil, enabling it to grow more innovative results. "We may use AI innovation to promote the inclusiveness of AI and let AI affect many industries more."

Xu Li said that artificial intelligence is the source of innovation, and the real subversive innovation comes from the brain hole of genius or the incredible conjecture and thought experiment. With artificial intelligence and supercomputers, will machines think? Will the machine guess? Xu Li said the answer is yes. In some areas machines have given humans very good examples, pushing the progress of human science in reverse.

What are the necessary conditions for machine conjecture? Xu Li said that first of all, with the development of computer computing power in the past 20 years, the demand for computing power of the best artificial intelligence algorithms has increased by nearly 1 million times in the past 10 years, which is a huge breakthrough.

"Theoretically, the more advanced the algorithm, the less computing power is needed, on the contrary, the larger the space we explore, the greater the computing power, which has nothing to do with data, because there are many algorithms that have relied on small data, or even do not need human data, but this exploration gives us the possibility of updating iterative cognition." Xu Li said that the artificial intelligence computing power device is the basic element to promote conjecture. The large device can be analogous to the particle collider, which knocks out new particles in random possibilities and explores the unknown world.

Second, because it is conjecture, many things are uncertain, so it is necessary to define the boundaries of industrial applications in the real sense or to define whether it is unreliable. In fact, there are many applications in human history that have been applied by conjecture before a suitable explanation can be given. For example, the Wright brothers did not know the principle when they invented the airplane, and even today they cannot explain the dynamics of the aircraft to take off with fluid mechanics, but this does not prevent the aircraft manufacturing company from creating a safe and controllable aircraft.

A large part of many artificial intelligence applications are based on such conjectures. For example, automatic parking in automatic driving, old drivers can often summarize the rules of parking, but these laws can not be converted to the computer in time. The computer itself has evolved a set of laws, and even scientists can't really explain the logic of computer reversing, but this does not prevent automatic parking from landing. Xu Li said that in the future, we can truly explore the boundaries of technology possibilities by iterating more scenarios.

Another, more complex application is decision-making in an open environment. For example, the application of intelligent transportation, usually in the solution of traffic light control, human beings have been difficult to give a perfect answer. "We're trying to solve this problem with artificial intelligence guessing on a very small scale, and we can save the average local waiting time in half, close to 20 minutes." Whether it can be extended to a larger scope still needs more exploration. But in this direction, Xu Li said that the machine conjecture has given us a surprise.

Xu Li said that another type of application may be commonly believed to be big data-driven applications, including smart cities, two-network construction, and smart communities. "But when we really enter these applications, we find that it is not so big data, and we are solving the daily low-frequency applications that are just needed."

For example, when solving the problem of falling objects from a high altitude, it is impossible to have too much data on falling objects. When solving the problem of community care such as fire and elderly fall, it is often zero data and small data, so the machine can only make possible guesses about these scenarios through general technology derivation and generalization.

Xu Li said that at present, in many cases, machines cannot have a complete control and iteration, so SenseTime adheres to the development of better artificial intelligence ethics and launches an ethical development concept of artificial intelligence based on controllable, green environmental protection, sustainable and people-oriented.

"What I would like to emphasize more is that we need to look at our AI ethical governance from a developmental perspective." Xu Li said that development has two meanings, first, to develop as the goal, all ethical governance is to make artificial intelligence better serve our society, better promote the development of society. "If you don't talk about goals, it will be very easy to meet any one of the governance conditions, but a single optimization will often make everyone get into trouble."

Second, the development of artificial intelligence is very fast, so it is necessary to balance different governance frameworks with the perspective of development and choose different governance policies at different stages. A negative example is that when britain built its first car in 1865, the United Kingdom introduced the red flag law, and a person waved a red flag 50 meters in front of the car, so the speed of the car could not exceed that of people. "Of course we have achieved complete safety and control, but the UK has also lost the dividend time of car development."

"So today, the conjecture of artificial intelligence and the conjecture of the machine may be the apple that Newton smashed on the head, and Shanghai's inclusiveness and openness is its best soil, enabling it to grow more innovative results." We may use the innovation of artificial intelligence to promote the universality of artificial intelligence, so that artificial intelligence can affect many industries more. Xu Li said.

Editor-in-Charge: Li Yuequn

Proofreader: Luan Meng