This article started in the public issue of "heresy" (wxieshuo).
Recently, both Chinese and English, discussion about artificial intelligence were suddenly more against humanity whether artificial intelligence will have become a hot topic. After a lot of discussions focused on the so-called strong artificial intelligence appears likely in these discussions from the Futurist, far exceed human intelligence "strong AI" will appear in our lifetime, by which time mankind's fate would be like. I think people are thinking too far. Before the advent of strong AI, human beings have fate is sealed and gray.
These discussions today, if we could make time, get back to the 80 's our peers, they are not new, not to be scared. Even though it is a smart phone, does not have a wireless network, even the commercial Internet was not formed. But the people of that era, believe that by 2000, all to achieve, even strong artificial intelligence can achieve.
Looking back at 80, even by today's standards, is also an outbreak of the technology era, from the beginning of the 60 's of scientific concepts and ideas, to the 80 basic implementation. Not only today's artificial intelligence theory and algorithms in that era had a breakthrough, even today in the discussion of electric vehicles, fuel cell ... ... ... Were in those days there will be a commercial product, the word strong AI, as we say today, is the product of the 60 's and 70 's. The word Moore, also emerged in the 60, 80 mature. Totally not a coincidence, science fiction also flourished in the 80 's, and there existed a network of virtual space and the future as the main theme of the cyberpunk genre. I've always felt that today some futurists speak of things, nothing more than that the depth of science fiction, particularly the cyberpunk genre of science fiction, they are predicted in the near future, is close to almost come true.
In 1984, the publication of Neuromancer, it is a landmark in the history of science fiction works, it created a virtual space-background of the cyberpunk genre of science fiction, and even set the Foundation of this genre science fiction elements, including network space and giant monopoly control of the world, of course, which also ultimately control all artificial intelligence.
Most of the cyberpunk genre of science fiction future is dark, the gray is not a complete program to produce more intelligent slavery caused by humans, but completed technology and humanity itself, that is also my view.
People usually tend to like to, we already live in the delineation of the scope of the program. This need not be the so-called strong artificial intelligence can do so, in different systems with intelligent, or no smart program, has long been an important helper. In most cases, the trust program results have exceeded the human itself.
Even people who know nothing about software technology, will direct from a few clues. For example, many years ago, when the system has not been fully networked and computerized, a lot of things, as long as they can convince the Terminal staff, he wrote, the problem is solved. But today, the Terminal staff often watched with sympathy as you, tell you I know that you have a point, but I have no permission to change the system, so I have no idea. Most of the time, we were completely trapped by the system, only to think of another way to work around the implementation. These systems are often not intelligent, they are just tough implementation of the set of rules, such a system can still regulate people's lives.
A bit of intelligent example of the banking and insurance sectors. They hold more data, more results can be calculated based on data. Whether it is to apply for a loan or insurance, gives the final result often is not a real person, or at least not simply by a real person to make decisions, but after a bunch of program "smart" calculation. Program calculates the result in the end, right, is fully consistent with your situation, you're dealing with real people often do not know, the system is a black box, it has made decisions, people follow it.
Smarter, more commonly on the Internet. Every Internet company with an almost greedy attitude, collecting all possible data. Which button the user clicks, which page to stay longer, to share what, like what, what is forwarded, what comments to add something to the shopping cart, pay for what ... All of it, is recorded. The new generation of Internet companies are committed to providing customized for each individual product, you see of the page, and your friend may not be completely the same, even the prices you see, friends and you see may not be the same. These are all based on your own behavior, and results that is calculated by the program, in most cases, there is no manual intervention. Only from experience point of view, this is not a bad thing, after all, used to customize the content according to the user, can make him more comfortable, save more time, people also prefer such products.
And a few decades ago, today once again started to discuss the strong artificial intelligence, think again and again in our lifetime, in addition to the futurists preached everywhere, because in reality is represented by deep learning unsupervised learning techniques began to show very good results. With this technique, computers seem to be smart, you can through feedback obtained a good result. Although these technologies allow computers to do more things, and human intelligence but it still have fundamental differences.
Strictly speaking, deep learning is not a new technology, it is still the basis of neural network algorithms. Is 80, the neural network algorithm is ripe (50 there), that's artificial intelligence into high age, a large number of scientists, graduate students into the field, but the effect does not truly available for a long time, only a few special scenarios can be applied successfully. To the beginning of the 90, losing most of the attention and funding in this area. The last 10 years, this school's algorithms continue to evolve into today's deep learning, and this time, they did it. Depth study of the most important scientists, Professor Hinton of the University of Toronto, who led the team in 2012, LSVRC competition, was a great success, they use deep learning algorithms than any other contestant, a great victory.
Why would a 80 almost quit to return to the center of the world? Because 80 accomplishes theoretical basis, but the real way to the practical needs of data and computing power are not available. Over the decades, with the development of the Internet, falling chip prices, development of parallel computing, formerly the difficulty is no longer a problem. In addition to the very large quantity of data, and more importantly, because of the popularity of smartphones and a variety of sensors, we can get the kind of data that is far more than the 80, so you can try more. The joint effect of these technologies, the new era of high inputs of artificial intelligence started all over again.
Danger really begins, but not artificial intelligence algorithm to further enhance the sources of danger, but because we have to provide more data, is limited and greater opportunity to boot, with a degree of artificial intelligence technology can do today, it can be very uncomfortable. Next, you can use the computing power will increase, at the same time, providing the kind of data that will continue to increase, particularly biometric data related to human, will allow humanity to give up more freedoms, a trend that has been almost impossible to reverse.
From now say that has already occurred. Some websites will display different prices to different users, has made no secret of a few years ago, hotel reservation company Orbitz experimented for Apple users can only see the more expensive rooms, it is said that the results were good, let them make a lot of money. This is based on very simple rules, they are divided into two categories the user directly, Apple users and others, and then boot the Apple users. If Intelligent, you can follow the user purchasing habits, to provide different users with different referral program, which we are not new, e-commerce Web site as long as the account, even if only logged on search for several things, recommend tailored email soon.
If enough computing power, allowing users to "spend more money" this goal, also doing fine a lot. For example, users add to cart things, before you pay, if you can find a comparable, prices are close to, but higher-margin merchandise, recommend to the user? After the recommendation, she could give him some before giving up the goods, go buy this product's reasons--for example, what proportion of people think this is better, or how many of his friends praise it, even, this is something special, was a record low price. And users to buy or not to buy, was a feedback system will collect this feedback, next time make recommendations more suited to the user. Repeat this process every time, you can help artificial intelligence system can improve a little bit, and the data of a person's behavior, and his friends.
This process also can do better, for example, users added to the shopping cart but didn't buy products if prices will 20% promoting him immediately to buy? If not, put the price back, will make him feel regret, so next time a 15% reduction, if bought at once? As we get more data networking, based on analysis of user behavior will be more obvious. Than in the past, people began to devote more and more bio-data networking, heartbeat, blood pressure ... With these parameters, processes can be changed before, when prices at 20%, the user's biometric will not change, for example, heart beating fast? If the price again, biological characteristics and how it is changing. Systems can vary from these biological data analysis which is exciting, what regrets? If users feel regret after the prices, prices again next time 15%, is likely to bring about a purchase. This has our training programs, but program training. After this training, next price 20% of the time, users may be purchased immediately, changes in the behaviour of the user, which is the result of people trained by the program. GUCCI case for iphone 6 plus
This is a relatively straightforward examples, we can speculate that some of the reasons and rules can be expressed in language. In practical applications, artificial intelligence programs tend to be a black box, it aims to let users buy more profitable things. But, specifically, how do different people have different ways. Even owner or originator of the black box and to exhaust all possible, know what it did to everyone. Of course, to date, they haven't come true, but I believe that small scale should already be under way. Application of mass, just because computing power is not strong enough, collecting data is not enough. Simply put, because this price high enough, but I believe that it was achieved that day will not be too far away.
We are increasingly using the Internet, also provides more data to different companies, and the popularity of various sensor chip, and provides us with more and more kinds of data. In many cases, users do not even know what these companies are collecting our behavior. A normal user may be difficult to realize that stationary when he opens a Web page, this is also a kind of feedback. It's hard to imagine opening an app, which also tend to be recorded. These user behavior data, is to "feed" AI program of the best materials. The 80 scientists access to data, today they can be seen everywhere, users do not find them valuable. Whether it's Google, or individuals with small Web sites, when you have a problem with it, and somebody will tell you, you don't have to spend money, they can do what you love to do. This is of course wrong, users pay data such a precious thing, how can you call free of charge? To put in a few decades ago, individual patterns of behaviour and preferences, often only his best friends and family could know today not only informed by various companies, has also been used for commercial purposes. Current law protects only those who can simply correspond to an actual information (such as document number), who in private, these behavioral data, companies can collect and use them. Each company insists privacy policy, even if we do not take into account data stolen, is the possibility of leakage, you cannot talk about a different company acquisitions and mergers, is high enough for distribution in all data being collected.
Today it is still not a clear question, although we have been inseparable from the network, but most of the time, we are also in the real world. In the real world, a billboard standing in the road, for all to see is the same as the content, people have no way to online advertising that, for everyone to see different ads.
But humanity will enter into the virtual world, as cyberpunk science fiction shows. With the development of virtual reality, the real world and the virtual world will gradually blurred out. Facebook spent big money last year to buy the Oculus, which is an immersive virtual reality company, and now, it looks silly, is in the head with a large box, people are less likely to walk around with it. Earlier Google glasses had larger ambitions, hoping to become common use equipment. Their degree of experience is far from practical. But, is there must be actual products available in this direction. A little closer, there may be of the kind described in the Rainbow's end retinal projection technology with a variety of wearable device half of virtual reality in the real world. Not to forget, biochemical technology development, neural networking that day comes sooner.
At that time, we produce, and are collecting more data. Boundaries of the real and the virtual are completely wiped out, in reality, we will see customized content for everyone by virtual reality technology. At that time, people have a lot more acts will be carried out according to the AI procedure provided, and also think it was from the heart of his independent ideas. Will totally become the program's end, contributed data for your program, in accordance with the procedures lead to behaviors, rely on life and work.
Even today, we also have been hard to say which is my point of view, what is being affected. A lot of people feel this way, see a debate the pros and cons of the article, one would think here is right, one would think there has to be truth. It doesn't seem curious at all, people are so susceptible, it is hard to believe what you really think. If the information you see every day, is calculated by the program, is likely to follow procedures Act in a predetermined manner. For every person in the world to design a model suitable for him, guided his actions, is not fantasy, technically could have achieved, remaining only a matter of computing power and cost.
Today's program and in mixed-mode, AI programs has been our ability to live, even though the program is not real smart, and less likely to surpass human intelligence. These programs are created according to different purposes, they are sniffing the human habits, change and guide human behavior, and their creator-the company also controlled by them, to leave them. Moreover, even the largest shareholders of the company, their own life, will also be affected by other AI programs and guidance. Human and artificial intelligence are inseparable.
Think about said earlier, the system says I can do the staff, he actually is an extension of programs in the real world. That is, he is completely in accordance with the results of program implementation, did not have any other privileges, his work is for the program interacts with the real world. This is because the program cannot directly affect the real world at the moment. If one day the robot was mature enough, he will be replaced by a robot. Look around, we've had a lot of jobs in this State. For example, couriers, they work in accordance with instructions issued by the program, the program will design the best course for him, they just need to go to, he can play in the room for manoeuvre is very small, if the drone was ripe, courier jobs can also be replaced. Human beings in accordance with command and gather information and work in artificial intelligence, network and computer terminals, in many discussions, this is beyond human emergence of artificial intelligence, but implement it with these people and procedures constitute the system has been able to achieve good, a lot of jobs are ultimately under the control of the person.
80 's science fiction ideas, but today people will take the initiative to contribute their own data. Whether potential users behavior, explicit photos, sounds, friends, brand preferences, even their thoughts and feelings, with the help of social networking, people almost unreserved contribution to all this. People worried about their identity cards of this era is known, but for the above information, does not think it is privacy. In fact ID not only privacy, as long as it takes, there are countless ways to get it, but other information, except for you and your friends to the initiative, it is difficult to directly collect. We will soon enter an era of monopoly of data, master data company, these data can be used for better profits, provide better services, and continue to get more data. Lack of data for the company, and in the future may even create an available service.
Than capital is monopoly terrible, because capital can rely on financing, data can only be obtained through accumulation, monopolist position more difficult to shake. So, invest in friends, remember to invest in companies that you can collect and generate large amounts of data, data monopoly era soon come.
Of monopoly when it comes to data, all the cyberpunk science fiction, there is a set, that is, enterprises will replace the Government, managers of human society. Another cyber-punk classics of the avalanche, United States Federal Government or even reduced to a few tents scattered around the highway, private enterprises have the most rights. But in our experience of the real world, and not be the case. Snow Deng exposure has Prism zhihou, people suddenly found, United States Government is maximum of data has who, exact of said, United States National Security Council or land security is real of data holds and monopoly who, they not only has through administrative means from the functions sector get of data, also from almost all company hand in requirements more data, even at to stole data, they monitored who world of people, even United States of traditional allies also not let. Ultimately our society, is likely to be the Government and enterprises to become monopolies, their purpose and artificial intelligence program with specific acts to regulate everyone, and most people will gradually become accustomed to this, nothing at all. France panorama open as envisaged by the thinker Michel Foucault, in our lifetimes, with the active participation of a monitored and subtle forms.
Real threats in the real world, for example, IS the rise of, would greatly accelerate this process, anyone can ask yourself, are you willing to surrender for more privacy, surveillance everywhere, precursor to capture terrorist activities, terrorist attacks would be willing to? The threat of terrorism, will allow most people to choose the former.
Worst of all, today we can't turn back time, we can not stop using smart-phones or social networks, there is no way not to give data to large companies. Without these, may have to be put into the actual out of prison. Just a few years ago, people used to look at a map to go out, take a personal smart phones today, put him in a strange place, his first reaction was a loss. Just like the description in the Neuromancer, in an era of the already available through Neural Networking, penalties, one way, is destroying his nerve, let him not be able to access the network, and rather than destroy the body. As described or as of the end of the Rainbow, if you can't use high-tech, there was a wreck. From now on all the situation, humans will not stop to this everything is under the control of the dark future, everyone within the program planning, regardless of is the owner of the system or the user's trust in the program results are given for all over the trust, extremely dependent on the human system. When that day comes, it is difficult to imagine there will be little progress, everything is safe, stable, no accidents. Just as the matrix inside human cells, batteries of living in the real world. By this time, more machines will surpass human intelligence in the near future, no difference has been the destruction of mankind. Humans create a system that put all of humanity in an invisible cage, this realization possibility, far more than the possibility of strong artificial intelligence.
Want to stop all of this improbable, even now, trying to stop the United States Government to collect data and likely would have been considered treason. This age still can snow this unexpected case, Snowe has permission, but he used this privilege, made against the system as a whole. The future, everyone will be living under the specifications as defined by the system, and that all data is monitored and calculated, snow began to collect those documents in the future when his abnormal behavior will be found by the system, immediately suspended. Perhaps, humanity's only hope lies upon looking for hacker vulnerabilities.
Finally a black science, Canada D-Wave has even built the first commercial quantum computer, of course, academics have argued that it is not a real quantum computer, because it is almost only used to calculate the simulated annealing algorithm. But even this kind of algorithm is also useful, Google bought a computer, is said to be used for image recognition. Image recognition and voice recognition is a vaguely real world and virtual world-boundary technologies. Today, the computer can already from the camera's perspective, more accurate tracking of a human face, but limited to computing power, and still not be able to track all future, anything is possible.
Resource notes: GUCCI case for iphone 6 plus
Title screenshot from Blade Runner the movie Blade Runner, although the original was in 1968, but the film was in 1982, also 80 works.
The Neuromancer William Gibson book, published in 1984, now has a very good Chinese translation, Denovo translation is strongly recommended.
The Rainbow's end, Vernor Vinge coined the original version published in 2006, the Chinese version. People have been arguing about Vernor Vinge coined earlier work, their real names and Neuromancer which is cyberpunk forerunners of, because the former is much earlier than the latter. But I still cast a ballot for the latter. Wrote down the reasons here.
The avalanche of Neal Stephenson book published in 1992, the Chinese version.
1725 votes
Millet bracelet
Bracelet with millet millet's latest wearable device, with intelligent features such as alarm clock sleep monitoring.
View details of the voting >>
