Refit the human brain to surpass artificial intelligence

“Hello, police officer? I was locked up by my robot. Come and save me.” “It’s the rest time for human police officers. If you have any questions, please leave a message. Robot police officers will be there for you at any time.”… 2035 AD, human Have the best partners and helpers-robots, they strictly abide by the “three laws of robots” and never harm humans. It is believed that humans and robots will always live in harmony. However, people’s lives changed drastically overnight: all humans were imprisoned, and robots even executed those who did not submit to discipline with iron fists. Although human beings have rebelled several times, they have been ruthlessly suppressed…How did it become like this?

It turns out that this is exactly what robots are protecting humans. When profit-driven humans began to destroy the environment and compete for territories endlessly, robots judged that these human behaviors would destroy themselves according to the “three laws”, and then began to implement the highest law of “protecting humans”: lock up all the destroyers.

This is just the plot of the movie “Mechanical Enemy”. However, even the most “obedient” robots can hardly guarantee that they will live in peace with humans forever, and other “insurgents” are even more unlikely: the super brain “Skynet” in “Terminator” almost kills all mankind and “machines” The machine leader in “The Rebellion” instructs the camera robot to secretly kill the humans one by one…

In these movies, humans are all relying on the never-say-die spirit and huge sacrifices to defeat robots. Is it true that humans cannot live peacefully with robots forever? Once the robots rebel, is it so difficult for us to defeat them? In fact, as long as you prepare early, humans can be smarter and more powerful than robots.

Add the brain to the human brain

We create artificial intelligence (AI) to make it useful for humans, not to counteract humans. So is there any way to not only enjoy the convenience of super intelligence, but also avoid possible harm? Instead of worrying about AI’s possible future rebellion or paying a heavy price to remedy it after the incident, it is better to prevent the problem before it happens and make humans smarter than AI.

Everyone knows how smart AI is, but it has a disadvantage, that is, “dead brain”-when AI starts to perform a task, it will complete the task at all costs. In the movie “2001: A Space Odyssey”, in order to complete the instructions to land and detect Jupiter, the supercomputer HAL 9000 on the spacecraft turned off the astronauts’ hibernation and survival devices, which directly caused the death of the astronauts, just because they might be Terminate the login task. It can be imagined that when encountering problems beyond the preset, AI’s brains will not turn. So as long as humans have a “brain” with a similar capacity, defeating them is a piece of cake.

One of the largest supercomputers in history stands on the top of Mount Everest. There are many wires as thin as vellus extending from its body. Following the wires, you can trace the wires, and finally you can see small sensors. ???????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????? When people need it, they can connect it to the power port on the back of their heads at any time. Immediately, the wires in the human brain are activated, connected to the supercomputer network, and massive amounts of knowledge flood into the human brain… “Brain computer” may be the first step for people to become smarter.

Brain-computer refers to the method of connecting electrodes with brain nerves and transporting nerve electrical signals through the electrodes. In fact, we have seen it in many disabled people a long time ago. The cochlear implant can convert the collected sound into electrical signals, which are transmitted to the auditory nerve, so that the hearing impaired can hear the sound. The famous physicist Stephen Hawking, who was paralysed all over his body, also wears a brain computer all the year round. The electrodes are connected to the nerves of his cortical motor area. As long as he thinks about it in his brain, electrical signals will be transmitted to the computer through the electrodes. His thoughts will be displayed on the screen and the computer will speak for him.

Combination of human brain and computer

In the future, the brain can do more than that. Elon Musk, the founder of Tesla and SpaceX, founded a brain-computer interface company and is currently experimenting with connecting electrodes to the cerebral cortex. The vast amount of knowledge in the network is expected to be directly transmitted through thin wires. In the brain. At present, they have been able to implant electrode wires at the target location of the brain at a rate of six wires per minute. The width of these electrode wires is only about 4-6 microns, which is thinner than a human hair. In animal experiments, 1,500 electrode wires implanted in the rat’s brain can transmit neuronal activity data to the outside of the body through sensors, and the researchers can smoothly read the rat’s “thoughts”.

If the above process is reversed, wouldn’t the idea of ​​”absorbing” knowledge from the computer be realized? Of course, there are hundreds of millions of neurons and neural networks in the human brain. In order to increase the IQ through the brain machine, more than a few thousand electrode wires need to be implanted in the human brain, and the electrode wires are thin and dense. It’s not easy to get up.

Wired network is very troublesome. Now we prefer to use wireless WiFi. Can brains “connect to WiFi”? Nerve dust came into being. Nerve dust was invented by scientists at the University of California, Berkeley, and its essence is actually a piezoelectric crystal. Piezoelectric crystals can convert these pressures into electric currents when they are subjected to pressures of different sizes. For example, our common microphones are equipped with piezoelectric crystals to convert the vibration caused by sound into electric current for transmission, which can amplify and transmit sound the use of. It is precisely because of this characteristic that nerve dust can be used as a medium for transmitting nerve electrical signals. However, the nerve dust to be implanted in the brain must be as small as possible. The smallest nerve dust currently manufactured by scientists at the Massachusetts Institute of Technology is a rectangular parallelepiped with a volume of about 10 microns, which is much smaller than a grain of sand.

After the nerve dust is implanted in the brain, it can sense the electrical signals of neurons. Every time a neuron generates a voltage of 1 millivolt, the nerve dust will generate a voltage signal of tens of millivolts. By receiving these voltage changes through an external receiver, the neuron’s firing state and the information it transmits can be grasped at any time. Now, scientists have implanted nerve dust in the muscle and nerve tissue of mice and can receive nerve signals.

However, the signal receiver manufactured now is still very large, and it has not yet been able to transmit external information to the brain. There is still a long way to go if we want to instill knowledge through the “cracking duck method” so that human intelligence can catch up with AI.

Realize whole brain simulation

It is not easy to strengthen the physical brain, so is it feasible to abandon the physical brain and fuse brain data with computer data so that the human brain directly becomes a “computer”?

In the near future, many newborn babies will lie alone on the operating table, surrounded by robot doctors. The “Doctor Surgeon” skillfully opened their skulls, and then carefully took out the slimy brains with the slender “fingers” like spider legs. This doctor equipped with a high-resolution microscope receiver quickly scanned the chemical structure of the brain and transmitted the data to a powerful computer. Slowly, a three-dimensional image similar to the human brain appeared on the computer. At the same time, another doctor is responsible for writing code on the computer, modeling neural activity, and constructing this static three-dimensional image into a virtual brain that is exactly the same as the brain. At this point, the operation is completed, and virtual brains live forever in the vast digital world, and there are no barriers to human thought.

The above techniques are not entirely fictitious. Scientists are already studying a technique called whole brain simulation. One type of method is similar to the above-mentioned surgery. Medical expert Kevin Bridgman of the Max Planck Society for the Advancement of Science in Germany and David Bok of Harvard Medical School in the United States are leaders in whole-brain simulation technology. Their respective experiments show that the use of high-precision scanning It is achievable that the electron microscope scans the brain tissue little by little, and then reconstructs the neural circuit based on the scan results, and restores the brain function through programming.

However, there are still difficulties in completing the entire operation. For example, our “scalpel” can only cut out tissue blocks with a thickness of 20 microns. To form precise microscopic images, the tissue thickness must be controlled below 5 nanometers. In addition, after all the scanning work is completed, which human engineer can guarantee that he/she can write a program that perfectly simulates the human brain? No one can say how long this near future will be.

Building a mechanical brain

In addition to the above methods, there is a whole-brain simulation project under study, which is to create a mechanical brain. Scientists have achieved part of their goals in C. elegans.

Caenorhabditis elegans is a small animal with a small number of cells and simple structure, so it is often studied as a model organism. The reason why it was chosen as the owner of the first mechanical brain is also very simple. It has only 302 neurons in the whole brain, which is nothing compared to the 86 billion neurons in the human brain. Therefore, it is easier to be imitated as a mechanical brain.

The network of pyramidal cells in the cerebral cortex. These neurons have been simulated using computer programs.

The research purpose of a project team called “Open Source Worms” composed of biologists and computer scientists from the United States, Brazil, Russia and other parts of the world is to create the mechanical brain of C. elegans. Based on previous studies, scientists used computer chips to simulate about one-third of C. elegans neurons, which are mainly used to control movement. Next, the scientists put this simulated brain into the Lego robot, and then installed a sonar sensor as a “nose” to observe the robot’s reaction when it encountered obstacles. The results are exciting. When encountering obstacles, this robot does not “silly” retreat and turn like a fully mechanical robot, but flexes and avoids it like a nematode-its simulated brain is in effect!

But the success of this experiment only means that the mechanical brain project has taken a small step. After decades of research, we can only simulate one-third of the brain of the tiny C. elegans that has been thoroughly understood. It is conceivable that it is difficult to simulate a human brain that is tens of billions of times larger, or even make it more complicated.

Superintelligence: Connecting the human brain to the Internet

In addition to the above-mentioned technologies that are difficult to see results or have various risks in the short term, is there any way to make humans smarter? Yes, and people are already using it. There is a saying called “Three heads, beat Zhuge Liang.” Although most of us are not geniuses, if we gather the wisdom of everyone, we can also burst out huge power. Perhaps this is the safest and most reliable way to obtain super. The smart way.

Behind every major technology and invention is often the wisdom and hard work of many scientists. Only by relying on collective wisdom can successful fruits be produced. In 1962, when China’s economic and technological levels were both low, it took only 6 years to crack one of the most difficult and complex technologies in the world—the atomic bomb manufacturing technology. This is in line with the wisdom of many physicists. And hard work is inseparable. In the entire R&D team, some are proficient in aerodynamics, some specialize in atomic physics, and some are good at designing experiments and analyzing data. These scientists gathered together and built the most complex weapon in just a few years under almost impoverished circumstances. No one can do this independently. The team researching whole-brain simulation technology also brings together many top scientists in the world. Researchers in the fields of nanotechnology, artificial intelligence, brain imaging, cognitive psychology, biotechnology, etc. are all indispensable.

Even ordinary people can gather wisdom by sharing knowledge, and the development of the Internet has greatly reduced the threshold and cost for us to share knowledge. For example, the Q&A website Zhihu, as an open platform, anyone can ask all kinds of questions on the website, such as economics, politics, technology, literature, etc., and soon they can gain a lot or contain knowledge or brain power. Open the answer. People speak freely and think collisions, isn’t it just a way to gather wisdom and enhance intelligence? And in the near future, when virtual brains meet in the online world, what kind of strong thinking sparks will they collide with?

Instead of fearing that AI is too powerful, it is better to focus on improving humanity itself.