Will brain-like artificial intelligence become the "new favorite" of the next generation of AI?

How do high-tech robots reflect the "high" in their names? The method is simple. Whether they can think like humans is the best way to judge their technological content. Of course, robots at this stage cannot think like humans, but their operating principles can be realized by simulating the way humans think, which is the so-called brain-like artificial intelligence.

Brain-like artificial intelligence has an information processing mechanism similar to that of the brain. It can collect and process information from different senses, make appropriate judgments on its own, and finally command the output system similar to motor neurons to perform actions similar to human reactions. In other words, the brain-like intelligence envisioned by humans must include the ability to process sensory information like the brain, and must also be able to manipulate highly complex electromechanical systems such as flexible joints and well-coordinated muscle groups. Although today's human technological level has begun to show its strength in these two aspects, it has not yet achieved disruptive achievements. Brain-like artificial intelligence that fully meets the above requirements may still have a long way to go.

The most difficult part of developing brain-like artificial intelligence is that we don’t even understand the brain itself, so how can we develop artificial intelligence by imitating the brain? Although we have a certain understanding of the physiological mechanism of signal transmission between neurons, and have also studied the physiological activities that each part of the cerebral cortex mainly controls or participates in, we know almost nothing about how the information transmission and analysis between the two is achieved.

In addition, another major difficulty in the development of brain-like artificial intelligence is that there is no effective mathematical model that imitates the operation mode of the cerebral cortex. On the one hand, this is because the physiological mechanism of the neural activity of the cerebral cortex is still unclear, and on the other hand, it is too difficult to establish a universal mathematical model of the cerebral cortex. First, we know very little about the brain, and second, we cannot abstract the working mode of the brain into a mathematical model that can be understood by computers. These are the two major reasons for the difficulty in the development of brain-like artificial intelligence.

The above article only talks about the difficulties, which may make people despair. In fact, humans have done some small work in the research of brain-like artificial intelligence. Corresponding to the difficulties, humans have made strenuous efforts in further studying the brain and modeling the working mechanism of the brain.

The gap between the microscopic (neurons) and the macroscopic (cortical brain regions) is called mesoscopic brain science in academia, and it is a gap that needs to be filled in the 21st century brain science research. Fortunately, there are already technical means such as optogenetics and two-photon microscopy that allow humans to regulate the activation and inhibition of specific neurons and even observe the collaboration of thousands of neurons at the same time.

With the continuous revelation of the working mechanism of the brain, humans have also made certain progress in the mathematical modeling of brain information processing. In recent years, artificial neural networks based on deep learning have continuously created new myths in the application of artificial intelligence. AlphaGo has achieved such a huge success based on this model. There is actually no need to fully understand the specific mechanism of this model, and this is not a problem that can be explained in a few words. We only need to understand the artificial neural network based on deep learning as a mathematical model that imitates the working principle of the brain. After all, only when the input information is digitized can the computer or AI process it, and the model is the core tool for digitizing information.

Visual recognition and speech recognition based on artificial neural networks can be said to be two application fields with rapid development, and have even replaced humans in many fields. A simple search in the browser will reveal many novel applications. Face recognition, expression recognition, gait recognition, specific target recognition (license plates, fruits, tumors, etc.) are no longer new. Unmanned driving road condition analysis, multi-language recognition and translation have also gradually become commercially available.

Here, we have to mention the domestic AI processor Cambrian. This processor is the world's first commercial deep learning processor. It is based on a self-developed AI instruction set and has completely independent intellectual property rights. It has better performance and energy efficiency than traditional general-purpose chips in key areas of AI technology such as visual recognition and speech recognition. The successful launch of the Cambrian chip indicates that China's brain-like AI research is completely at the forefront of the world.

However, even the most advanced artificial neural network models currently available are terribly simple in front of a real brain. Humans have only borrowed a few of the most superficial operating modes of the brain to achieve such great success. The bright future of brain-like artificial intelligence research is still worth looking forward to.

Will humans one day be dominated by high-tech robots powered by artificial intelligence?

Before humans fully elucidate the brain's operating mechanism and establish a highly universal brain information processing model, the development of brain-like artificial intelligence will be completely controlled by humans. As for whether artificial intelligence with self-awareness will be born and whether humans are at risk of being enslaved, such grand propositions should be left to science fiction directors to discuss!