According to statistics, the most important and promising technologies in the field of science and technology in recent years - big data, machine learning and artificial intelligence topped the list.
Today, even if you do not know what artificial intelligence is, you must have heard the great name of AI (Artificial Intelligence), and even chat about personal insights. If you are a heavy user of various information platforms, then you must have deeply experienced the accuracy and magic of big data push.
What the author wants to popularize today is not machine learning and artificial intelligence, but to connect the other end of that technology, to know and understand the mechanism of our own human brain in turn through the study of artificial intelligence, and then to explain what consciousness is from a scientific point of view.
Our thoughts come from the brain, yet we cannot directly observe the brain's operation, so it is very difficult to understand the working mechanism of the brain.
To unravel the mysteries of the brain, let's first look at how computers work.
As we all know, a computer/computer is composed of five basic parts: input, output, control, operation, and storage. The function of a computer is summarized as follows: to take the input information, go through a series of operations, and then output a result of the operations.
Why is a computer called a computer? Because its basic working principle is the same as the human brain: we transmit information to the brain through vision, hearing, smell, taste, and touch (input); then the brain analyzes and judges the information in a series (operation); and finally makes the corresponding response (output).
Where the input and output parts are intuitive, the computing part (the brain) is a black box where no one can see exactly how the brain processes information.
Since we can't see how our brains analyze information, let's take a look at how computers do their calculations.
Anyone who knows a little bit about programming knows that the earliest computer arithmetic programs were made up of many if else instructions for logical operations. It probably looked like this.
if input 1 then output 1
else if input 2 then output 2
Although such a program is also capable of handling increasingly complex logic, and even many complex programs far exceed the analytical capacity of the human brain, such a program is soulless.
Because all logical judgments are programmed, the computation process is fixed and flexibility depends only on how many if else branches are written in your program.
But the human brain obviously does not follow a fixed logic to process information, the soul of the human brain from learning, the so-called live and learn, with learning and cognitive changes, the human brain's judgment of the same information is constantly changing.
Computer technology has developed to this period, although the arithmetic power is improving, but in the end is still only a machine that runs according to instructions, not yet talking about intelligence.
However, once computer technology enters the artificial intelligence stage, things start to get interesting.
The first impression many people may have of AI is that of intelligent-like decisions simulated by a large number of operations. In fact, no, the qualitative change of artificial intelligence is to allow the computer to achieve the learning ability of the human brain, so that the computer can also learn to constantly acquire and update knowledge, rather than simply running instructions.
The ability to learn is the key to achieving intelligence.
For example, today's common application of face recognition technology, it is not just to identify the static ID photo only, no matter what kind of expression you do, or from different angles, different distances, it can be accurately identified, so how does the computer do it?
For example, big data push can accurately push the content you are interested in to you, and how does this work? Is it possible that computers can already read and understand human language?
Instead of discussing the details of machine learning, we will introduce an important concept in this paper: the Neural Network model.
Neural networks are a model for implementing Machine Learning, a network structure inspired by the physiological structure of our brain - interconnected neurons.
In other words, the neural network system is the theoretical basis for implementing machine learning, which is the same as the principle of learning in our human brain, only there are some differences in the specific implementation.
The artificial neural network algorithm for machine learning consists of three network layers: the input layer (input information), the hidden layer (analytical calculations), and the output layer (output results).
Input layer: is the data feature input layer, the number of input data features corresponds to the number of neurons of the network.
Hidden layer: the middle layer of the network, the number of hidden layers can be 0 or many layers, its role is to accept the previous layer of network output as the current input value, and calculate the output of the current result to the next layer. The hidden layer is the key to the performance of the neural network and usually consists of neurons containing activation functions to further process the features at higher levels of abstraction to enhance the nonlinear representation of the network. The number of hidden network layers directly affects the model fitting effect.
Output layer: The network layer for the final result output. The number of neurons in the output layer represents the number of classification labels. (Note: When doing binary classification, the number of neurons in the output layer is 1 if the activation function of the output layer is sigmoid, and 2 if the softmax classifier is used)
The point of introducing neural networks is that the mechanism by which our human brain processes information is also a network of neurons. Each neuron has an input, a processing, and an output. Then many neurons are connected together to form a network that processes and reacts to various signals, which is the root of how the brain can think.
Receiving area: dendrites to the cytosolic part, there is a change of potential, stepwise electricity generation. Stepwise means that the dendrites receive different sources of synapses, and if the more sources received, the greater the effect on the cytosolic membrane potential and vice versa. And the information received is integrated within the cytosol.
Trigger zone: a potential integrated in the cell body that determines whether or not the onset of a neural signal is generated. Located where the axon and the cytosol meet. It is also the part of the axonal thalamus.
Conduction zone: the part of the axon that, when generating a nerve signal, obeys the all-or-nothing law to decide whether to conduct the nerve signal down.
Output area: The purpose of the nerve signal is to allow the nerve endings, the neurotransmitter or electrical signal of the synapse, to be output in order to affect the next receiving cell, called synaptic transmission.
With neural networks, is it possible to do decision analysis? -- Not yet.
As a newborn baby, although it already has a brain, it cannot think yet. There needs to be a long learning process to slowly build up cognition and logic.
So, how do neural networks acquire cognition through learning?
The basic principle is not complicated, it is to find out the law from a large amount of historical data and build a function/algorithm according to the law. This algorithm is not fixed, but in the process of learning data constantly adjust the accuracy, the more data learned the more accurate this algorithm. And this algorithm in the form of neural network inside the existence of the interwoven network structure .
That is, the learning process of the human brain/neural network is a process of training to form neuronal organization structures through a large amount of empirical data.
Our brain's cognition is trained by data, and this cognition is an objective one in the form of a neural network (hardware rather than software).
The brain's thinking process is not a nebulous, random act of consciousness, but the processing of signals by a network of neurons, an objective entity in response.
At this point we have a general understanding of how our brain learns and thinks about decisions, not as a purely conscious activity, but as an objective network of neurons. This is a brain hardware structure rather than software, and it is important to recognize this.
Only by recognizing this can we in turn analyze our brains by looking at artificial neural networks, because the two actually work on the same principle.
For the sake of textual brevity, in the following, if neural networks are mentioned, they refer to artificial neural networks/computers when not specifically stated; if they refer to the human brain, they refer to the network of neurons in the human brain. This is used to distinguish computer neural networks from human brain neural networks.
Let's use the simplest example to analyze how to use neural network training to "brainwash" a computer.
For example, we want the neural network to learn to recognize what a cat is.
We first feed the computer a large number of pictures of cats, and the neural network will gradually find commonalities based on these pictures, extract features, adjust the weight of this parameter, and gradually train an algorithm (network structure) with a high recognition rate.
For example, this algorithm might identify a cat based on two ears, two eyes, a head, four legs, and a tail as features. Even more precise parameters, what shape the ears are, what is the approximate proportion of the head to the whole, etc.
Thus, the computer managed to learn how to identify a cat by learning from a large amount of data.
It looks like a successful learning case.
But with a neural network trained this way, how would the computer determine if we showed it a picture of a dog? -- it might think it's a cat, too.
Because the dog also has two ears, a head, four legs, and a tail, it matches the characteristics of the cat learned by the computer.
If we tell the computer that this is a dog, not a cat, will the computer immediately change its perception of cats and dogs?
--No. Because the computer's judgment of a cat is based on the complex algorithmic path of a neural network, which is trained over time by learning and does not change completely with a single correction; therefore, even though you tell it a fact (that it is a dog), it will not accept it immediately.
You see, we "brainwashed" the computer to recognize the dog as a cat, even if you tell it that it is a dog, it will not "believe" you.
Why does this happen? Because when we fed the computer data to learn, we only fed it data about cats and did not provide data used for comparative classification, such as dogs, rabbits, and other mammals.
This example allows us to conclude that feeding the neural network/brain with a large amount of one-sided information and filtering out other information used for comparison and classification will train the neural network/brain into a one-sided cognitive model, which will misjudge other types of information.
--Through information filtering, cognition can be instilled.
Consider the following example again.
Suppose we input 10,000 pictures of cats into the computer learning process, but only 1,000 of these 10,000 pictures of cats are real cats, and the other 9,000 are actually dogs.
And what will be the result?
Obviously, the neural network will think that the dog is the cat (90% likely to be the cat) and the cat is not the cat (only 10% likely to be the cat) after learning. If you give it a picture of a dog and tell it: this is a dog, or give it a picture of a cat and tell it: this is a cat. The computer will think you are wrong.
With this example we can conclude that as long as the neural network/brain is fed with the wrong information a sufficient number of times, this wrong information will eventually overwrite the correct information. Instead, the correct information will be judged to be wrong.
-- A lie repeated 1,000 times becomes the truth.
Similarly, if we feed a computer neural network with 10,000 pictures of a cat, but tell the computer that it is not a cat.
Obviously, this will train the computer to perceive that "cats are not cats".
-- As long as we repeatedly instill the determination of a negative fact, eventually the neural network will assume that the fact is not true.
Through these examples above we can intuitively see that the basic principle of brainwashing strategy is similar, which is to train the neural network by repeatedly instilling certain specific information in a large amount, and eventually let the neural network form a fixed determination path, once this determination is formed, any information of different determination (even if it is a fact), will be denied by the neural network.
This is what we often call inertial thinking, values, ideology.
See, through the analysis of neural networks, we are able to intuitively understand how ideologies are formed.
Now that we understand the basic principles of brainwashing, how can we tell if we are being brainwashed?
Before getting into this problem, we need to understand an important mechanism in deep learning of neural networks: the validation mechanism.
As stated above, a neural network is a complex network path/algorithm that is gradually built up through learning, and this algorithm is not fixed, but gradually improved with learning. The validation mechanism is the important mechanism used to improve the algorithm.
For example, if our computer has developed a network/algorithm for recognizing cats through learning over time, how do we know if the algorithm is good enough? It is through continuous validation.
For example, if we input a new image and the computer calculates it through the network, if the output is correct, then we get a positive feedback that the current algorithm is correct.
However, if the computer calculates the output incorrectly (e.g., recognizing a dog as a cat), then a negative feedback is obtained. More importantly, the computer then needs to calculate the error and then adjust the existing neural network structure according to the error. This process of adjusting the algorithm may consume a lot of arithmetic power.
When a neural network examines a message, if this message matches the current path/algorithm, then it gets a positive feedback with an error of 0 and passes easily.
But if this message does not match the current path/algorithm, it will get a negative feedback, and this time it needs to calculate the error and adjust the algorithm, which is quite arithmetic intensive.
Similarly, our brain's neuronal network receives information with ease and pleasure if the information matches the current neural pathway and the brain gets a positive feedback.
But if this information does not match the current neural pathway, the brain will get a negative feedback, and at this time, the brain will start to check what exactly went wrong? What is the cause of the error? The brain then feeds the neurons to re-correct the neural network based on the results of the analysis. This process of correcting the neuronal network consumes a lot of energy and effort, so the brain will have a very uncomfortable feeling (similar to how our muscles feel when we work out).
It is important to understand the brain's feedback mechanism for information. This mechanism can help us to explain many phenomena.
For example, why we have fun when we play and feel pain when we study; for example, why the victims of Stockholm syndrome will instead defend the perpetrators. And so on many abstract and even unconscionable phenomena.
We can also use this mechanism to test whether our brains have been "brainwashed".
The reader may ask, "Shouldn't the criterion for determining whether one is brainwashed be whether one knows the "truth" or not? Why is it necessary to understand the brain's feedback mechanism for information?
Because, truth is a hard to identify and even somewhat subjective concept.
Everyone thinks that what they know is the truth, but in fact no one can know the whole truth. The amount of information contained in the whole truth is so huge and intricate that it may be far more than what each of our brains can store; just like an iceberg, everyone can see only the tip of the iceberg, no one can see the whole iceberg, and if the truth is an ant on the iceberg? Then you will never find it, I'm afraid.
As a result, most people are under the illusion that "What I've learned is the truth, and you've all been brainwashed." -- When we think this way, we may already be brainwashed.
Therefore, identifying brainwashing by the brain's response is a more direct and effective method.
We all have this experience, when we brush the shake, immersed in one video after another, the mood is incomparable pleasure, excitement, moved, a sense of identity, see the excitement and even a rapid heartbeat, heart pounding. So keep brushing the video, can't stop.
Judging from the brain feedback mechanism, this is because the content of these videos is highly compatible with the neural network in our brain, so our brain is constantly receiving positive feedback, feeling pleasure and excitement.
And this is the time to stop and reflect on whether we have been brainwashed by such content?
What's worse, Jitterbug also has its own neural network, and its neural network will keep pushing you similar content (big data push) based on its learning of your preferences; this in turn makes this kind of content take up more and more weight in your brain, which also makes your brainwashing state more and more serious.
On the other hand, there are times when we occasionally brush up on one or two not-so-common videos. These kinds of videos can make you feel very uninteresting, boring, ridiculous, and even disturbing. You will feel too painful to watch such videos and want to turn it off immediately.
Judging from the brain feedback mechanism, this is because the content of these videos contradicts the neural network in our brain, and our brain is subject to negative feedback and needs to readjust the neural network, so we feel discomfort.
This is also the time to stop and reflect on whether we have been brainwashed by other different types of content.
Now that we understand how to recognize brainwashing, what can be done to combat it?
First, let's correct a misconception about brainwashing and anti-brainwashing - we all think that people who are brainwashed are deep in it because they don't know the truth, and as soon as they are exposed to it, they will come out of it.
Actually, it is not like that.
I believe that everyone has had the experience of being surrounded by people who are deeply involved in pyramid schemes or online shopping, and no matter how much you reason with them about the facts, they will not believe. Why is this so?
Because the brain's judgment of information is based on a neuronal network, which is a very complex structure physically present in our brain, this structure is not immediately altered by exposure to the "truth".
Therefore, the root of the fight against brainwashing is how to adjust the neural network structure that exists in the brain.
Let's start with a short story.
There is a paragraph, if you want to make a southerners and a northerners quarrel, the easiest way is to discuss this issue: bean curd brain with salt or sugar? Are scrambled eggs with tomatoes salty or sweet?
The author is a northerner, grew up eating salty scrambled eggs with tomatoes. 18 years old, admitted to a college in the south, I remember the first time in the school cafeteria to eat scrambled eggs with tomatoes, a bite almost did not spit out, how is this sweet? Too bad!
But six months later, I no longer find the sweet-tasting scrambled eggs with tomatoes difficult to eat, but I think the taste is very good. What's even more amazing is that when I go back home to the north, I can still eat scrambled eggs with salty tomatoes and they are still just as delicious.
From then on, whether scrambled tomatoes and eggs are salty or sweet is no longer a matter of debate for the author.
This short story contains the simple logic of brainwashing, anti-brainwashing, right and wrong, and truth.
Which is right, salt or sugar in scrambled tomato eggs? Which is the truth?
The author initially felt that the sweet tomato scrambled eggs were too hard to eat, really because the sweet tomato scrambled eggs are harder to eat than the salty tomato scrambled eggs?
The answer should be clear to all readers, in fact, the right or wrong and the truth about scrambled tomatoes and eggs depends entirely on the author's habits. When the author is accustomed to both flavors, the salty and sweet ones become both right.
When our brain's neural network leans one way, we become exclusive of the others; if we can get our brain to accept the others, we unlock new skills, and as we unlock more skills, we can deal with more complex conflicts.
The author remembers that many years ago talk shows usually had two guests, one guest speaking on the pro side of the argument and the other guest speaking on the con side. I don't know when this format of media programs seems to be less and less available.
Today, I sometimes listen to what is being broadcast on the "News Feed", and every time, people around will cast a strange look, as if to say, what age, there are still people listening to the "News Feed"?
Understanding how brainwashing works and how to identify it, we have a strategy to combat it.
As we begin to realize that the neural networks in our brain are leaning more and more in one direction (brainwashed), we should try to expose our brain to information in other directions, even though at first we may feel that information in other directions is wrong and absurd. It's like the author's first time eating sweet-tasting scrambled eggs with tomatoes.
For example, if you are an atheist, you can try to get in touch with religious knowledge, and vice versa.
For example, if you are a conservative, you might try listening to a liberal point of view and vice versa.
For example, if you are a big fan of certain people or things, you should listen to what people who are against them are saying, and vice versa.
For example, when you feel like you're getting a little negative in your quest for the "truth", you might want to get a little baptism of "Newswire".
Improving and enhancing the neural network structure in our brain by receiving information from different aspects is one of the most direct ways to fight brainwashing.
And a further approach is to build logic, abstract models and methodologies.
For example, in the example above, if we want the computer to learn to recognize cats, we show the computer a lot of pictures of cats, and if we want the computer to learn to recognize dogs, we show the computer a lot of pictures of dogs. However, there are millions of animals on the planet, and the time and energy consumed by this method of learning can be imagined.
So, what are the ways to improve the efficiency of learning? Yes, there is classification modeling. For example, we can identify tigers, leopards, lions, cats, lynx, etc. faster by categorizing and extracting features from felines. -- this is abstract categorization.
For example, every boy as a child has a martial arts dream, always fantasize that they can become the book the martial arts warrior. But after learning physics and Newton's laws, this martial arts dream basically to be destroyed, because according to Newton's second and third laws, the law of gravity, we know that the so-called light body like a swallow, left foot on the right foot right foot on the left foot of light power is impossible to exist. -- That's the logic.
For example, when learning multiplication, we only need to memorize the 9-9 multiplication table from 1*1 to 9*9. We don't need to write down all combinations of numbers greater than 10, and we can also multiply two-digit, three-digit, and multi-digit numbers. Because we do multiplication by the rounding algorithm, not by writing down all the permutations. -- that's the methodology.
If we build the right and sufficiently complex neural networks of abstraction, classification, logic, and method in our brains, we can greatly improve the efficiency of learning information and reduce the number of brainwashed routines.
Why do we need to understand the principles and methods of brainwashing and anti-brainwashing?
Because we have entered an era of highly developed information technology, we are floating in the flood of information almost every day. We should learn how not to let information sway our thoughts and moods; learn how to use information to solve real-life problems.
It is more important to know how to solve problems based on information than to complain about reality; it is more important to use the knowledge to improve the situation than to dive into the "truth".
When humans think, God laughs.
Consciousness, one of the most mysterious areas in this world. Especially when we start to think deeply about "who I am", the brain will gradually enter a state of near madness, as if the computer in solving a nested algorithm accidentally entered a dead loop.
And by looking at artificial intelligence, we can in turn understand how our brains think.
Interestingly, the neural network formed by the neurons organized in our brain physically determines our thinking characteristics; we then pass our thinking characteristics to the neural network of big data through various Internet platforms today; the neural network of big data then feeds these characteristics to countless other human users through its algorithms.
Thus, an extremely tiny neuron, composed of a brain neural network, and then interconnected through the network of big data, as if we all end up on top of a huge neural network, which is the Earth civilization.
Discover beauty, own beauty, enjoy beauty
Let's build our beautiful life together
Let's build our beautiful world together