Current location - Education and Training Encyclopedia - University rankings - Fresh science: building digital people with you
Fresh science: building digital people with you
Ray kurzweil, the most famous American futurist in the world, predicted that in 2045, human beings will be able to upload their brain consciousness to computers and realize digital immortality.

This concept seems very abstract. What does this mean? For example, in the future, people will be able to scan your brain in sufficient detail to construct a digital "you". When you die, the digital "you" begins to take over the real you and live in a simulated computer environment. Digital Elysium has your cat and TV, and the surrounding environment is exactly the same as before. Your relatives can call you as before. "You" is like living in a dream, not realizing that you are just a digital virtual character. In this way, human beings have achieved digital immortality.

Kurzweil is the president of Singularity University. He has received nine honorary doctorates and two presidential honorary awards. As we all know, he is also a famous prophet. However, will this prediction be successful in 2045?

Understand the human brain

If we want to create an identical person, a person with the same memory, personality and consciousness as before, we must first create an identical brain and establish a neural network exactly like the real you, and all this must start from simulating the neurons, the most basic unit of the neural network.

Now people can simulate human neural networks on microchips, using the principle of changing the synapses between neurons responsible for transmitting information. For example, you give an artificial intelligence neural network a task and let it try again and again. Every time it behaves well, you give it a reward signal. Based on some simple learning rules, the size of each synapse will gradually change, and the connections between different neurons will also change. With the passage of time, the neural network gradually formed.

The news that Google AlphaGo defeated the top human Go player may shock you. AlphaGo's super intelligence comes from the neural network of deep learning. So, can this technology be extended to copy the human brain?

At present, it seems that things are far from simple. There are about 1000 billion neurons in the human brain, and each neuron can initiate about 1000 connections, which means that it has about 1000 synapses that can store information. Multiplying these about 1000 connections by about 1000 billion neurons is obviously amazing. Some people may say that after the emergence of quantum computers, these are not problems. Then, let's further assume that even if humans really have the ability to create 654.38 billion+000 billion artificial neural networks, how can the uniqueness of the connections between neurons in everyone's brain be replicated?

There is no ready-made scanner to find out the connection mode (connection group) of each person's neurons. Magnetic resonance imaging scanners can only achieve the resolution of 1 mm, while synapses are only a few microns. Of course, maybe scientists can kill you and slice your brain under a microscope. Then scientists can try to track dendrites, axons and their synaptic connections. This technology is terrible, but even so, the current technology is far from this level.

The brain structure is not that simple.

However, now that scientists have simulated a part of the connectome of mouse neurons, maybe in the future we can really scan your brain and extract the complete connectome. So, if you create a neural network with extremely complex neurons, will things be done?

Don't get too excited for a while, because copying a human brain is far from as simple as we thought. If you want to copy your brain, you need to copy its quirks and complexity, which determines your unique way of thinking. Any mistake in these details may lead to epilepsy, hallucinations, delusions, depression and anxiety. Therefore, it is not enough to copy your own connection group. If your scan can only determine which neuron is connected to which neuron, and reconstruct this pattern in the computer, you just create a disabled digital person.

Why do you say that? This is because the real brain is very unique. In a real brain, every neuron is different. For a simple example, some neurons have thick insulated axons like "cables", which transmit information at a very fast speed; Other neurons have thinner axons and send signals more slowly; Some neurons don't even send signals. All these neurons will have different changes at different times.

The way of signal transmission is also very different. There are hundreds of different synapses in the brain. Different synapses use different neurotransmitters, which will determine which neurons will receive these signals next. There are also some neurons that can "talk" directly without going through the synapse between them. Other neurons behave like a gland. Instead of sending precise signals to specific target neurons, they release a chemical soup that can affect a large part of the brain for a long time.

Artificial neural network can not simulate the complex structure of real human brain. The neurons in the artificial neural network are all the same, which will only affect their connection with the change of synaptic strength, ignoring the complexity of the biological structure of the brain.

Therefore, in order to copy a real human brain, scientists need to find out what kind of synapses are in each brain, what kind of neurotransmitters are emitted by synapses, and how fast the neurotransmitters are synthesized and absorbed. So, is this impossible? Maybe, but it sounds like the high technology of the future, not the technology just around the corner.

Are digital people conscious?

Even if we take a time machine to the future and can accurately copy the future of your brain, there is another obstacle. Is this brain really conscious, or is it a digital computer that imitates your behavior?

About consciousness, people have put forward dozens of influential scientific theories, each of which is different, and no one knows which one is correct. But in neurological theory, people think that computers can simulate conscious brains. If we can scan a person's brain in detail and simulate the structure on the computer, then the simulated digital brain will be conscious, and it will have memory, personality, emotion and primitive intelligence.

However, this does not mean that we are out of danger. Let's look at another situation. If one day you are kidnapped, there is only one window every day, and water and food will appear regularly to keep you alive. There is nothing in your room. The only thing you can see for decades is a vacant room of about 10 square meter. what do you think? Unreal or scary, in short, it feels unnatural, right?

Turn this room into a virtual world. If there is nothing in this virtual world, what is even more frightening is that you find that you don't even have a body. At this time, even if you are conscious, I'm afraid you want this consciousness to disappear as soon as possible. Our cognition and emotional experience are often related to the surrounding environment, and this relationship between brain function and the surrounding world is sometimes called "concrete cognition". So the next task is to simulate the actual body and a real world, and then embed the simulated brain in it. In today's video games, the real body can't be simulated. The body in the game does not have the normal elasticity or mobility of human muscles and skin. However, compared with simulating the brain, these technologies are already a piece of cake.

Are you still you?

So, now let's make one last hypothesis. In the future, we already have all the technologies. When you are dying, we will scan your details and quickly create a digital copy of you. Then, when you die, this simulated "you" wakes up with the same memory and personality as you. The virtual world in which "you" live seems not perfect. The taste may be different, the details may be lost, but it's good. You find yourself in a familiar world. This may be a simulated new york city, with the same dead people, but no rats or dust. Whether in the countryside or on the beach, the annual system upgrade will make the environment more real again and again. Unless the operating system crashes, "you" are not aging, injured or dead.

In short, congratulations: you have finally achieved eternal life! You can continue to communicate with the world while you are alive. You can continue to update Weibo and WeChat friends circle with your smartphone or email. Maybe "you" still have a job in the real world, as a lecturer or a director or a comedy writer, and pass on your wisdom to generations. It's like you went to another universe, but you can still get in touch with the world you once lived in.

For your friends and family, you still exist, only during the long journey. But is this you? Did you cheat death, or did you just replace yourself with a creepy copy? Obviously, even the most accurate simulation will not be you. The real you, with your body dying, will be a weird fake. But you in the virtual world don't think so. Just like you before you die, he will firmly believe that this is the real you.

If you created a digital "you" before your death, this problem is really difficult to figure out. Are you one person or two people, or is there a real you and a fake you in the world? However, before you understand this problem, you should understand that with the current technology, Kurzweil's prediction that digitalization will make people immortal in 2045 is unlikely to come true.