Speaking of it, this self-driving bicycle is promising. It is led by a team with a multidisciplinary background (including brain science, computer science, microelectronics, etc.). ) In Tsinghua University, and it was independently developed six years later. The core of this driverless bicycle lies in the research and development of general artificial intelligence chips. This chip named Tianji has experienced the integration of computing science and neuroscience, optimized the computing architecture and algorithm, and combined with advanced chip technology. Finally realized, as the first demonstration of the universal artificial chip, it appeared on the cover of Nature magazine this week.

Artificial intelligence, which we often hear, is a new technical discipline aimed at simulating and expanding human intelligence. Today, artificial intelligence is experiencing explosive development. At present, there are two main branches of artificial intelligence: one is based on the development of computing science, and the other is based on the research and development of neuroscience. The former has the advantages of big data but the accuracy is insufficient, while the latter has high accuracy but limited data. The platforms used by the two systems are different and incompatible, which greatly limits the development of general artificial intelligence. As a new interdisciplinary subject in brain science, computational neuroscience is committed to combining the advantages of two independent disciplines to balance data and accuracy. This new discipline will play a key role in the development of brains such as computing and artificial intelligence.

At present, the mainstream artificial intelligence imitates the human brain rather than the human brain to solve problems, so most artificial intelligence chips on the market are custom chips, such as Intel Xeon platform and Google TPU. Although these products have good performance, they are still close to traditional computing and can only be needed in specific scenarios. Scientists realize that the cognition and autonomous decision-making of human brain system is a very complicated process, and the ultimate goal of artificial intelligence is to fully simulate the human brain's functional perception environment, think independently and produce corresponding behaviors. In order to be different from the current mainstream artificial intelligence, a new concept-generalized artificial intelligence has been introduced into the field of artificial intelligence.

The successful demonstration of celestial motion reflects the possibility of the fusion of various neural networks, algorithms and models, which is a major breakthrough in the field of general artificial intelligence. This is a big step towards universal artificial intelligence. I believe that the realization of general intelligence in the future will make SIRI smarter, robots smarter and driverless more reliable.