Will Artificial Intelligence Really Destroy Humanity?
Artificial intelligence is fundamentally different from the human brain. It is essentially just a program. After it has developed to a critical point, it is possible for it to learn, update itself, and develop itself, leading to an increase in its IQ. The speed is far greater than the evolution speed of human brain IQ.
In a more intuitive description, the IQ development curve of artificial intelligence may be an exponential curve, that is, before a certain critical point, it is very, very flat. After passing the critical point, it will slowly start to accelerate, and the speed will become more and more Fast, until a nearly vertical line.
People treat the growth of artificial intelligence as if they are waiting for a train. At the beginning, they feel that they grow very slowly. Until one day watching this train slowly approaching, slowly approaching, and then, swishing, passing you away, never stopping to wait. That is to say, the time for them to be at the same level as human IQ may be very short, and after surpassing humans, they will quickly leave humans behind.
One day after that, the IQ of artificial intelligence will be far higher than that of human beings. On that day, their IQ is much higher than that of humans, and they can already be regarded as different species. Even though they are not malicious to humans, they belong to a higher existence, and they act based on their own interests. A choice that could still lead to the extinction of mankind.
For example, everyone will understand. Take the relationship between us humans and endangered animals. When we make decisions, it is impossible to consider their interests. Sometimes we do not intentionally want to harm them, but it is Our human inadvertent actions have led to their mass demise, such as excessive deforestation, massive emission of poisonous gas, oil spills, and so on.
So their inference is that artificial intelligence is a time bomb. Once their IQ grows beyond the scope of our human understanding, they will lose control, and our human beings will be on the verge of being destroyed.
This theory sounds reasonable at first glance, but when you look deeper, it is far from the case.
First of all, it is very difficult for artificial intelligence to master the real learning abilit.
I want to popularize popular science knowledge for everyone. If artificial intelligence wants to achieve its goal of growing through learning, its precondition is that it can update its own code program by itself, that is to say, it can write programs by itself, and then Use the written program to replace the program that you run. At this stage, all the so-called artificial intelligence, machine learning and other applications do not have such functions. At the present stage, the so-called machine learning, deep learning, artificial intelligence, etc., their so-called learning is only to modify their own behavior patterns through a large amount of data. For example, judge the user's preferences through some characteristic behaviors of the user, judge the future weather through the collection of weather data, and judge which move should be taken now through the past chess games. In other words, the so-called learning can only modify their behavior within the limits of his program. To put it bluntly, it is the chess program. He can only play chess. It will recommend news, programs that predict the weather, and can only predict the weather. So this so-called learning is not real learning.
If you want to have a real learning ability, you must have the ability to rewrite your own running program, so that you can add and modify the functions of your own program according to the needs of the external situation to adapt yourself to this world, and then you can achieve real growth. Is it possible for artificial intelligence to reach such a state one day? Maybe there is, maybe not, but no matter how to achieve this state, it is very difficult.
This requires computer scientists to write a code that can develop programs, and this code that can develop programs, the programs he develops also need to be able to develop programs, this logic must continue, and go on infinitely. The developed program is still capable of developing the program.... This requirement is extremely difficult.
Specifically, it is very easy to develop a program that generates specific behaviors according to specific needs; and it is possible to develop a program that can develop different programs according to different needs, although it is difficult; but the current requirements Yes, this program can be developed in an infinite loop, which is almost impossible, at least at this stage. This is a bit like recursion in code development, but recursion in computers can be solved by stacks, queues, loops, etc., and this is the business logic of the real world. Although it is program development, it is not an internal matter of program development, but The program develops external business logic, so it cannot be solved recursively. And it seems to be an infinite loop, but in fact it will always be a finite loop, otherwise it will exceed the computer stack limit and crash. Developing a program that can develop a program is an infinite loop, a real infinite loop.
Although the logic is very convoluted, the reason is very simple. It is very, very difficult for artificial intelligence to master the real learning ability
People who are good at thinking may refute that the creativity and learning ability of human beings do not seem to need to be realized through their own production, so why does artificial intelligence need to do this? In fact, you are wrong. The learning process of human beings is actually the process of constantly updating their brains. Every time a person makes a little progress and learns a new skill, his brain will undergo corresponding new changes, because the brain’s execution code will change. They renew themselves, you just don't realize it. This is the natural mechanism of our brain, and it is also where we humans are truly different, that is, only we humans can actively evolve ourselves through continuous learning.
The code executed by the computer is dead, and once it starts running, it is fixed. It is impossible for him to continuously update his code during the running process. Therefore, if he wants to realize self-evolution, he must be able to develop his own program to replace, updates itself, which creates the circular dependency restriction mentioned above.
Second, Cognitive limitation
What is it that causes a kind of higher life to suppress other lower life, that is, what causes this so-called higher life. To clarify this issue, we can infer this conclusion by comparing two aspects.
Let’s first look at the difference between humans and animals. The physical difference in brain volume is obvious, and the resulting IQ gap is inevitable. Although brain volume is not strictly proportional to IQ, the overall correlation is very strong. But is that the only difference, is it just your IQ that makes you superior? Those children who were raised by wolves and monkeys (note: such cases have been reported many times, you can search for them if you are interested), have the same brain capacity as humans, but they do not have all the basic abilities of humans. Is it more advanced?
You could say they don't have the IQ of a normal person because they haven't been brought up properly. In fact, you are wrong. They just missed the critical period of language and cannot communicate normally. It is not that they have no IQ at all. After systematic training, they can still have an IQ that is lower than normal people, but definitely higher than that of animals. What is the difference between such a person and a normal human being, if not IQ?
If you are not convinced and still do not accept the comparison between normal humans and them, what about a normal 10-year-old child? Have their brain development been basically completed, and their IQs are exactly the same as or even higher than normal human IQs? Granted this comparison, do you think it's okay to treat a 10-year-old child the same way as an adult? Are adults supposed to be higher than what children are supposed to be? Even children grow into adults after a few years.
If you don't agree, then imagine a world where everyone can only grow up to the state of mind of our 10-year-old children, and then stop developing. Can such a world have scientific progress? Are we not superior to them? But their brains and IQ are exactly the same as ours.
Therefore, the gap in IQ is only the basis for the difference between high and low species, not the essence. So what is this essence? It is the difference in cognitive level caused by learning! referred to as cognitive differences.
To sum up, if human beings develop a brand-new life form by themselves, and can improve their IQ and cognitive level through learning, then because it is based on the progress of human civilization, it is impossible for it to surpass the existing human beings.
Thanks for Reading