Wallfacer: I suggest that humans be bad

Chapter 17 The first step in brain-inspired intelligence

On June 15, 6, in Shuguang Building, Chen Xi, who was racking his brains to create in the office, heard a dull knock on the door.

He casually scratched out a few paragraphs of waste in his notebook, then put on his hearing aid, "Come in, please."

Ms. Say came in and put a United Nations document on the table, "Sir, they hope that you can use Sugon Technology's games, movies, and TV series to spread the values ​​of human civilization's victory over the three-body civilization and victory."

"What if I disagree?"

"They said the army would be sent to take over Dawn."

Chen Xi grinned helplessly, and said, "A group of robbers, well, let me think about it..."

"In this way, the script of Wandering Earth 20: In order to intercept the Trisolaran fleet's pursuit, the coalition government restarted the artificial intelligence MOSS, and used all the power of the world to build [-] star-class warships, and [-] soldiers volunteered to die to stop it The fleet of the Trisolaran civilization ended up with it in the end, and won miserably."

If it becomes a reality, it is estimated that there will be two thousand rings for the new version of the Spring Festival.

"Wandering Earth Four, without the hindrance of sophons, hundreds of years of development, two rounds of technological explosions, human science has achieved incomparably brilliant results, and the latest warship even has the fantastic speed of light navigation technology, so the coalition government decided to return. Take back the homeland on which mankind depends.”

If it were real, the Trisolarans would have caught up long ago and used super weapons, such as singularity bombs, to blow humans and the earth together.

"Wandering [-] mainly has two lines. The first is the earth's turn, and the second is that the light speed fleet sets off first to start the war."

"I'm making a TV series, talking about the cultural changes of human wandering for hundreds of years, and inserting some ETO elements to destroy the old-fashioned plot. For games, you can engage in repairs, defense of planetary engines, or space wars with the Trisolaran civilization. "

"Old rules, personal heroism is not allowed, there must be elements of a coalition government..."

Chen Xi talked a lot in bits and pieces.

Say wrote down one by one, and then said: "The second thing is that the analysis camera has arrived, and it has been placed in the computer department according to your request."

"Well, I see.

After thinking for a while, Chen Xi said again: "Ms. Say, can I trouble you to find Dr. Ding Yi here, I heard that he has recently turned to theoretical physics research again, trying to find a way to get rid of it in high-energy particle experiments." The method of sophon interference."

"Yes, but there are very few people who support his research, because theoretical physics research consumes too much resources, and it is impossible to get results now."

"That's a good relationship, so let's say that Suguang Technology is willing to support his research, although it doesn't have much resources."

……

After getting the analysis camera, Chen Xi happily started research: using chips and algorithms to simulate the movement of neurons.

With the assistance of the camera, people can see all the movements of the neurons. It took Wei Cheng and his team more than two years to figure out the neuron movement algorithm.

Based on structure, neurons can be roughly divided into four categories: unipolar, multipolar, bipolar, and pseudo-unipolar neurons.

Chen Xi asked the researchers in this way: First, the neuron chip must be able to perfectly simulate neurons with four structures, and the structure can be switched according to needs, so that future brain-like hosts can deploy different and required simulated neurons, and can run perfectly matching motion algorithm.

Second, the chip must be small enough to fit into a thin smartphone, and to be able to perform remote operations with the highest authority on various smart electronic devices.

The second requirement is easy to say, but also difficult to say. The simplicity lies in the fact that technical problems are easy to solve, and the difficulty lies in that it is not easy to pass the approval of governments due to privacy and other issues.

But compared to artificial intelligence, this is a small problem.

The combination of the two requirements stumped the engineers under Chen Xi: the chip is small, and the memory must not be that big.

Even when the main body of the four motion algorithms is not running, the amount of data is no less than that of the operating system of a smartphone. Once it is running, the generated data can fill a hard drive with hundreds of GB in half an hour.

Chen Xi hated iron and scolded: "Stupid, I asked you to study a new generation of data transmission technology, distributed computing technology, and cloud storage technology, isn't it for this reason?"

"The buildings in the future must be full of chips, some are responsible for storage and some are responsible for computing. Computing and storage are separated, and neuron chips can be simply bound to data transmission."

In the 17th year of the Era of Crisis, neuron chips were born. The small ones are only as big as the fingernail of a little finger, and the big ones are as big as a palm. The former is installed on small electronic devices, and the latter is the basic unit of supercomputers and neuron master brains. structure.

Fortunately, Keiko Shanshan was still working on the algorithm, and when she learned that Chen Xi had the chip, she used it directly.

……

In a certain experimental park of Sugon Technology, the brain-inspired learning experiment has begun.

Volunteers put on a metal cover on their heads. Inside the metal cover are cryogenic liquid helium and superconducting coils. Simply put, it is a nuclear magnetic resonance instrument, but it is far more advanced than that.

Afterwards, he began to write and analyze the movement of neurons in the volunteers' brains captured by the camera at the millisecond level. After the movement information is processed by a computer, it can form a neuron network model that can be analyzed and studied by brain scientists.

Due to the problem of computing power, it took a whole week for the computer to process the information captured by the camera. After that, the brain scientists input the movement information of the neurons into the brain-like brain bit by bit.

This process took a month, but it still failed because the brain is too small, with only 100 million simulated neurons, while the human brain has nearly 1000 billion neurons.

No way, they can only guess which neuron movements are the movements that occur when writing, and eliminate them bit by bit.

This is an incomparably huge workload. Chen Xi didn't want to waste his limited life, so he chose to hibernate for two years.

In 18, Suguang Brain-like controlled the robotic arm and wrote a line crookedly.

Brain scientists and computer scientists realize that as long as the neuron movement information corresponding to the behavior is accurately input into the brain-like body, and the translation system is good enough, the brain-like body can perfectly reproduce the behavior corresponding to the neuron movement. It does everything, no matter how complex.

The translation system is software that converts the analog neural signals of the simulated neurons into digital signals that the machine can understand.

With a slight upgrade of this thing, there is an application that has no practical value for the time being—real artificial intelligence: the analysis camera captures the neuron activity of the wearer, converts it into a digital signal and sends it to the machine, and the machine responds.

It has no practical value because the analysis camera is too big and cumbersome, and the translator only records the translation sentences of the neuron activity of one of the 80 billion people when writing certain words, which is not universal.

In 19, a major breakthrough was made in computer technology, and the computing power was [-] times stronger than before. Traditional silicon-based chips have come to an end, but there are still photonic computers in their infancy.

Countries are building a new generation of supercomputers one after another, and Sugon Technology is the same. Countries see the potential of brain-like intelligence and have provided a lot of resources, technology and manpower to support it.

(End of this chapter)