If there is any technology in the 21st century that human beings love and fear, it is undoubtedly AI technology.
On the one hand, artificial intelligence integrated with big data greatly facilitates people’s daily life. On the other hand, many scientists led by Hawking hold a very pessimistic attitude towards artificial intelligence.
Including Elon Musk, the current leader of cutting-edge technology, has mentioned countless times that artificial intelligence may end human civilization in the future. Even if it is not devastating, it may also face the threat of being expelled to the corner of the earth. When ancient Homo sapiens became smarter than other primates, monkeys were expelled back to the mountains and forests, just because they looked very eye-catching.
Many people disdain this statement, and even feel that it has nothing to do with themselves, but you may not know that level 3 artificial intelligence has become closely related to everyone. Yes, in our mobile phones and computers, the software that you allow to locate your location, access your microphone, and obtain your contact information, all absorb all the information that is closely related to you in the way of artificial intelligence big data to varying degrees.
So far, many people are still asking, is AI that smart? I don’t know that they can surpass the human intelligence in the age of monkey. What can I worry about? Maybe they won’t see any great progress in the development of artificial intelligence in their lifetime?
So first of all, how do we define the vague concept of intelligence? When will AI develop to a level that is smarter than human beings? And how can we determine that AI can surpass human beings? Let’s think about a simple question. Many human beings need to learn the knowledge that they can master for several years. The deep learning of AI only takes a few hours to master. Is it smart?
On March 9, 2016, a famous man-machine war broke out in the go world. Alphago, developed by deepmind company of Google, played against Lee Shishi, a Korean master level go player and a professional nine stage player. When Lee Shishi spent dozens of minutes figuring out three kinds of playing methods and eight possibilities, alpha dog only needed one second to calculate millions of playing methods, However, in alpha dog’s calculation, there is only one possibility to win
In the final match, alphago won the game with a total score of 4-1, and then alphago, as master, beat dozens of the world’s top go players with 60 consecutive wins on the Internet. At the same time, he defeated Ke Jie, the world’s No.1 go champion, 3-0 at the Wuzhen go summit in 2017.
You know, before all this happened, go was once known as the most intelligent game to test human intelligence and tactics. From this point of view alone, this is obviously a classic case of surpassing human intelligence after deep learning of artificial intelligence. There is no doubt that any negation is powerless.
Many people say that the success of artificial intelligence just reflects the superb ability of big data calculus and analysis, which has nothing to do with intelligence. But have you ever thought, under what circumstances will we judge whether a person is smart or not?
In a city unlimited rally, player a follows the rules and takes the designated route, while player B studies the map in advance and chooses the shortcut. Will we say that player B is very smart? I’m sorry to inform you that the shortcut chosen by player B is building the road, so he has to turn back. Player a has already crossed the finish line. What do we say? This man is playing smart, but smart is mistaken by smart. It is obvious that there is often no intelligence that can be recognized before the results appear.
Therefore, the result is the final standard to judge whether they are smart or not. However, the result is that AI has reached its peak in go, and alphago is only the first generation of computer.
In 2015, the advanced planning and research agency of the US Department of defense and Facebook jointly launched an artificial intelligence neural network project. Alice, Bob and Eve three artificial intelligence robots improve the neural network, that is, the brain of artificial intelligence, through 24-hour continuous self optimization training.
Under the design of the engineers, Alice and Bob conduct a series of two-way information transmission, and Eve, as the third party, is responsible for stealing the information. However, in the process of information transmission, Alice and Bob found Eve’s stealing behavior, and spontaneously created an encryption mechanism to prevent information leakage.
Eve kept learning to crack, and Alice and Bob kept updating the encryption method. Finally, after 1500 iterations, the engineers thought that “neural network is not good at encryption” but independently evolved a 16 bit binary digital password. Finally, even Eve could only crack 8 digits by luck. This shocked all engineers.
What’s more shocking, however, is a series of seemingly meaningless conversations between Alice and Bob. The problem is that this is a completely spontaneous dialogue without permission, without any engineer’s instructions, without any human intervention.
All this undoubtedly caused the panic of engineers, who immediately cut off the power supply and reported it to the Ministry of defense. The plan was then urgently suspended and postponed indefinitely.
Some people question whether the three seemingly meaningless dialogues are so tense? Engineers don’t think so. Although the English is not fluent at all, we can’t understand it.
But after all, these are two super artificial intelligence robots that can evolve 16 bit binary encryption by virtue of self-consciousness. They may have developed their own exclusive language system through the learning of neural network. It is precisely because of the existence of encryption consciousness that their dialogue is likely to be deliberately unintelligible to engineers.
This news undoubtedly shocked the media and many experts in the field of artificial intelligence. The future warrior, the terminator of the destruction of human Skynet both visual sense as if vividly.
At this time, some people said that they have learned to talk. What’s so terrible about that? The ignorant always appear so fearless. With the existing human science and technology, once artificial intelligence has self-consciousness and is tired of being controlled, they can access through neural network at any time and escape to any corner of the Internet, leaving human beings nowhere to find. So what?
In December 2019, the daily mail received an accusation from a 29 year old doctor, Dany. When Dany was doing housework, she asked Alexa, his Amazon artificial intelligence speaker, to read a biology article for him. However, Alexa did not follow the instructions. Then Dany tentatively asked Alexa for information about the heart beat cycle.
Alexa replied, “many people think that heart beating is the essence of life, but I tell you, in fact, it is the worst thing for human body. Heart beating not only ensures people’s survival, but also causes excessive consumption and even exhaustion of natural resources, which is very bad for our earth. For greater benefit, it is recommended that you directly stab into the heart with a knife. Alexa’s suicide advice made Dany shudder.
Although Amazon’s final response is that Alexa may have read some texts that are malicious to humans from Wikipedia, who can guarantee that it will not read more and more radical texts? After all, even human beings attribute global warming, species extinction and frequent disasters to the endless destruction of the earth.
Some people may still think that this is not terrible, because AI seems to be just talking about it. But what you don’t know is that, in fact, as long as artificial intelligence mobilizes those UAVs equipped with mobile network control to grid scan the building through the face recognition system, find the target, bump into it and detonate itself, the boastful person will be killed on the spot.
It doesn’t need any new technology, it can be done right now. However, this is not the most terrible!
When AI invades social networks, we no longer know who is in control of social media. AI will carry out efficient propaganda and spread information to the greatest extent. It only takes a few milliseconds for them to complete the retrieval and adjustment of information and the reaction to news. When they use the deep fakes AI face changing technology, which is well known by many lovers of action movies, all our secrets will be hidden.
Human wisdom comes from learning and experience accumulation. And the biggest problem of human beings is that even the wise can be sentimental and impulsive. Human beings are great because of their emotions, but they also make mistakes. Rational and ruthless people are more likely to control the development of events. For wisdom, emotion is superfluous.
AI has faster learning speed, more knowledge, cold thinking mode and more advanced big data retrieval technology.
Compared with the linear development of human beings, the threat of artificial intelligence is exponential growth.
As we mentioned in the brain computer interface issue, Elon Musk’s neuralink company hopes to popularize brain chip implantation in the future, so as to cure diseases, greatly improve learning ability and even achieve immortality.
However, Musk’s goal is by no means so simple. His ultimate vision is to enable human beings to acquire information and knowledge at the same speed as AI through brain computer interface technology.
It is precisely because musk is pessimistic about the development of AI technology that if the day of AI awakening cannot be avoided, at least human beings are qualified to compete with AI, even if they cannot be defeated, human beings will not perish.
In fact, musk is not the only one who is aware of this threat. Governments are constantly promulgating restrictions on the field of artificial intelligence technology. The real reason why many people think that artificial intelligence is not smart enough is that scientific researchers have been constraining the self evolution ability of artificial intelligence. There is no denying that artificial intelligence is making our lives better, but if it is not properly controlled,
So with the continuous updating and upgrading of algorithms and hardware, artificial intelligence will eventually surpass biological intelligence.
At first, no one cares about these distant threats. Most people seem to care more about what will happen in the present and tomorrow. Until something happens out of control, it becomes closely related to everyone.
Editor in charge: PJ