My research question is, how we can benefit from Artificial Intelligence without causing any dangers. The reasons that I am interested in this research question are that, firstly, I watch several movies and series about technology. The main problem that comes up with technology is always about Artificial Intelligence, such as in ‘I, Robot’ and ‘Person of Interest’. They all talked about when artificial intelligence has a huge improvement, accompanied by some disastrous problems. Secondly, I like video games. Games nowadays will let us decide what might happen next depending on what we choose. ‘Infamous Second Son’, a video game I liked, the decision I made can let the game’s ending become evil or good. Artificial Intelligence means machines or robots have their own thoughts, which also include superintelligence (machines which can perform the best current human thoughts in a large domains). From Siri to self-driving cars, artificial intelligence is now transforming its landscape to mountains of industries. It is invading humanity’s life bit by bit. In my opinion, I think almost all people in the world might care about the question, especially industries associated with Artificial intelligence. 

From a problem-solving program to today’s Artificial Intelligence, AI has come a long way. In the early development of Artificial Intelligence, humans simply just put commands in the box and it gave us the same thing we gave to it. In other words, we get what we put in. After that, humans unveiled machine learning. We get what they respond. The reasons why AI could be powerful are, firstly, it has a fast speed. “Signals can travel in computer at the speed of light while Neurons propagate 100 meters per second”. (Bostrom) Secondly, the size of computer can be even larger than a house while human’s brain can only stay in the skull. For human, what we think about intelligence is that there are two extreme value (Smart to Fool). However, for AI, it starts at the lowest point, and it develops his intelligence from zero to mouse-level intelligence, then chimpanzee-level. And after lots of years’ hard working, it will get to human-level intelligence. But it will not stop, it will be infinitely developing. Sooner, it will become even smarter than humans. 

As we know, there are mountains of advantage of Artificial Intelligence. Firstly, the development of technology opened a new era of communication, education, and science. Machines can now recognize and translate speech even in a noisy area, such as Siri. the appearance of self-driving cars proved that artificial intelligence has reached a new height of automation. Mercedes-Benz, a global car manufacturer, has unveiled the autonomous bus in the Netherlands and it successfully completed a 12-mile course without breaking any driving laws. Secondly, Artificial Intelligence might be the only way to prevent cybercrime. While human enjoy the scientific and technological achievement, the rapid growth of technology also created a new kind of criminal: cyber criminals. According to Intel Security Group, more than 400 billion dollars have been lost in the global economy because of cybercrime. More importantly, our personal information would be easily accessible through the internet. The process of taking down cybercrime is too hard for humans. However, it is easy for AI because the process need prediction, classification and control of complex system. “Interestingly enough, the solution may be too advanced for humans but doable for something created by humans: Artificial Intelligence. The rate at which digital crime occurs leaves little hope for human intervention to prevail without human error. AI has a much better chance at detecting and analyzing appropriate defenses against cybercrime because it, too, is a product of technology.” (Tatera, 2) “Watson”, IBM’s AI computer, became well-known as the winner of Jeopardy!, an American television show featuring quiz competition in 2008 by defeating two of the most successful players on the show. IBM used Watson to apply algorithms in order to watch for and identify any new threats. Thirdly, A.I. could improve the efficiency for health care and economics. “Companies are leveraging AI to transform health care, specifically in the areas of precision medicine, genetic testing, diagnostics and therapies.” (Guzowski, 3) The latest new shows that Artificial intelligence could be used to identify skin cancer. As for economics, AI could be used for solving some complex problems, such as data for demand, production, equilibrium prices and quantities. They can all be directly observed instead of calculating. Since economists can compute point prediction and theoretical models (Noussair, 82), it could also evaluate the changes. And it would be possible that AI could create an ideal economic network if the technique is mature and let the AI has access to data. In the future, Artificial Intelligence would play an important role on human life.

However, there is no denying that Artificial Intelligence might cause some dangers if we teach it in wrong ways. “Before the prospect of an intelligence explosion, we humans are like small children playing with a bomb. Such is the mismatch between the power of our plaything and the immaturity of our product.” (Bostrom) Humans have always wanted the machines matching humans in general intelligence since the invention of the computer in 1943. When humans could actually let the dream come true, we began to fear. We were afraid that machines would replace humans. 

There is a myth about King Midas, who always wished that all the things he touched would become gold. One day, he actually had the power. Everything he touched became gold. However, when he touched his lovely daughter, she turned to gold. The myth is not only talking about greed, but also as an illustration of things we create that are too powerful to control. Firstly, Artificial Intelligence could eliminate many jobs which would cause the unemployment rate to increase, such as farmers and taxi drivers. They can all be replaced by AI Robots. From CNBC Technology, Artificial Intelligence would eliminate six percent of jobs in US in five years. Secondly, Google’s Alpha Go AI defeated the world’s best Go (a kind of board game) player. It was just a kind of AI which only good at one area, Go. However, what if AI can master thousands of things in different area. That is actually happened and it is called superintelligence. “If it starts acquiring resources to accomplish its goals: The seemingly harmless chess goal therefore motivates harmful activities like breaking into computers and robbing banks.” (Muller, 3) As the Muller mentioned in “Editorial: Risks of Artificial Intelligence”, there is a little chance that superintelligence would rather be extremely good or extremely bad. Even when there is only a one percent chance the plane we take will crash, we still think that it is too risky. AI might be evil because they have some values that are contrary to humans. Think about if we tell the Artificial Intelligence to make humans laugh. At first, it might do some amusing action to gain the laugh. If Artificial Intelligence find a more effective way by putting electrode into the facial muscles and make humans laugh forever, it would be a disastrous thing. Although it was exaggerated, but some people believed that if the superintelligence became dangerous and uncontrollable, it might cause the extinction of humans. Thirdly, what if some well-funded nongovernment organization created automatic killing machine such as ‘The Terminator’ in order to control the world. It would also cause a lot of trouble to countries and citizens. 

“However smart a technology may become, it is still a tool to serve human purposes” (Etzioni, 3)

We all know that what Artificial Intelligence could bring to us. Consequently, we need the solution to the question of how we can benefit from Artificial Intelligence without causing any risks. Different people might have different opinions. Some people said that we should never develop the thing anymore since we all know it is dangerous, while others held the opposite opinion that we could actually let the AI grow without any limitation. Lots of people simply believed that putting the AI into a box is a good solution, but how do you make sure that it will not get out by himself because even human hackers can find bugs all the time. Even if we can control it at the beginning, we could not keep it forever. Sooner or later, it will get out, because is is not possible for a creature to control another creature which is much smarter than it.  For instance, can chimpanzees keep humans in a box? The answer is definitely no. As a result, the solution is actually very hard to find. However, there are two reasonable solutions. 

The first one is we can use an AI guardian to guard AI. In other words, let one kind of AI control another. “All societies throughout history have had oversight systems. Workers have supervisors; businesses have accountants; schoolteachers have principals. That is, all these systems have hierarchies in the sense that the first line operators are subject to oversight by a second layer and are expected to respond to corrective signals from the overseers.” (Etzioni, 3) From the article “Designing AI Systems that Obey Our Laws and Values.” We knew that there are four kinds of AI guardians. The first one is Interrogators. There are some questions human could not give the definite answer, where Interrogator AI would show up. The second one is the Monitor. We need a kind AI to detect whether the AI would break the law or not. For example, a self-driving car might violate the speed limit because it would to get the destination as soon as possible. The Enforcer, as the third one, helps enforce and build a flawless law about AI. The forth one is the Ethic bots, which in my opinion, are the most important. AI would not only obey the laws, but the sense of our values as well. If AI guardian can control AI well, there is no such a risk as the extinction due to AI. However, the solution also caused another question: Who will guard the AI guardian? Humans must have the final say about it. 

The other way was that when we create and teach the AI machine, we also need to teach them a sense of our values. (Grady) Then, if it gets out, we will not panic since it will be on our side and share our values. It is different comparing to the Ethic bots because we therefore do not need the AI guardian. They can figure out what is good and bad by themselves. It has no difference to a human who is well-trained. For example, if a child did not get well-educated and his parents did not even care about him, he has a big chance to become a rascal. If the child is well-educated, he will have few chance to behave badly. 

For how to solve the risks of AI, different answers deal with different questionings. A superintelligence system needs substantial and subtle training which is hard for an individual to operating a superintelligence. As Grady Booch said, comparing to super doing, super knowing is very different. Superintelligence might not be a problem because it cannot do thing, it only can think. After all, all of these AI should have on and off switch. So, when the nightmare really come true, we can simply unplug it. 

“We are on an incredible journey of coevolution with our machines. The humans we are today are not the humans we will be then.” (Bostrom) We do not need to care about what bad consequence it would bring to us. It actually would be a big problem since there are still lots of issues we have to attend nowadays, such as how do we let the Artificial Intelligence be well accepted. The question that people might ask is, “Should we fear it?” Now, every new technology brings with it some query. When we first saw cars, people believed that we would cause the destruction of the family. When we first saw telephones, people were worried it would destroy face-to-face conversations. At a point, we saw the written word become pervasive, people even thought we would lose our ability to memorize. Indeed, these things are all true to a degree, but these technologies can also bring to us things that extended the human experience in some profound ways. (Booch) Actually, if we are afraid of what would happen if superintelligence loses control, we might never have big progress on developing AI. As Bostrom said: “I don’t think just ringing the bell and screaming is actually what the world needs, it needs to get on with the actual hard work”

In conclusion, the answer to how we can benefit from Artificial Intelligence without causing any risks is that human should teach the AI humans’ sense of value when we build and teach it.  From my perspective, humans need to make sure all the Artificial Intelligence have the on and off switch. We need to make sure if things go wrong in the future, we can stop it immediately. As the Nick Bostrom said, “This to me looks like a thing that is well worth doing and I can imagine that if things turn out okay, that people a million years from now look back at this century and it might well be that they say that the one thing we did that really mattered was to get this thing right.” There is still a long way to get through all these problems. Until today, AI has already brought to us too much surprise than we expected. How many benefits we can gain from Artificial Intelligence in the future is unknown. And we are just at the beginning now. 
