As a novice programmer and a participant in Lego robotics I find the controversy surrounding artificial intelligence very intriguing. Programmers, computer scientists, and researchers alike have debated about the possibility of artificial intelligence becoming more intelligent than humans. Because I do have some knowledge of how computers work I can see why this topic is sparking so much interest. The thought of something that we created having the potential to surpass us is riveting.
It’s impossible to fathom the idea that humans may lose their spot as the alphas of the world. In this paper I will break down the arguments surrounding this topic by putting them into simpler terms and prove why one side may be superior to the other. Artificial Intelligence is defined as the study of man-made computational devices and systems which can be made to act in a manner which we would be inclined to call intelligent. John McCarthy, a retired Stanford professor, coined the term “artificial intelligence in 1955.
In 1950 Alan Turing published “Computing Machinery and Intelligence” and that led the way to the Artificial Intelligence eventually becoming a field of its own. Artificial intelligence, otherwise known as AI, can be found in many forms such as robots, computers, chat bots, and GPS systems. When digital computers were first introduced back in 1937 they were only as smart as the humans that programmed them, meaning they had no way of improving themselves. The way artificial intelligence has become smart is through the algorithms it has been programmed with.
An algorithm is defined as a procedure or formula for solving a problem. Programmers have begun programming computers with algorithms that allow for multiple variables and the computer has to make a decision that best suits the variable it’s given. In recent years programmers have enabled computers to store information gathered from the decision it made and use that information for another time. To put that in simpler terms the computer is learning from its own decisions. This function can be seen in programs called “Chatboxes. For example, Sim Simi is a chatbox; its programmers first programmed it with a response to a set amount of variables. For example, if the user inputs “Hello! How are you? ” it decides between “Hello! I’m doing well” or “Hello! I’m not doing well. ” When introduced to a new variable the program saves it and responds. At first the response may seem inadequate but at the program encounters that variable more it adapts and is able to answer in a better manner. Not all computer systems are created equally, each have a different amount of computing power.
The lowest are Constant Action Systems which the computation power is very limited and the system can only do a few things such as perform simple mathematical problems (Omunhundro). We have already surpassed that level and we are currently progressing towards Meta-Reasoning Systems which allow for modeling of certain aspects and upgrades them; that leads to an improved version called Self-Improving Systems that can “completely redesign itself” (Omunhundro). The most advanced level are, Fully Rational Systems, which is the level of artificial intelligence that is has created controversy.
With this amount of processing power the systems can then begin to think and begin to evolve at a faster rate. These types of programs have increased in recent years and continue to grow, but not in a controlled and steady manner. The growth of Artificial Intelligence can be graphed using an exponential growth graph. The point at which the graph begins to curve upward at a faster rate is called the singularity point (Kurzweil 3). “There will be no distinction, post-singularity, between human and machine or between physical and virtual reality” (Kurzweil 2).
Picture a world with androids walking among us but they look like us and act in identical manners. With every discovery and breakthrough in this field we are shortening the amount of time it would normally take to make such progress. Researchers and scientists will make twenty years of progress in only fourteen and then make fourteen years of progress within seven years (Kurzweil 5). With this amount of rapid growth it is easy to see how artificial intelligence can become something that can be abused. One of the problems with artificial intelligence is that in the wrong hands it can be used to destroy (Boden).
Robots are able to think and receive messages quicker than humans and if a robotic soldier is implemented into warfare the results can be devastating. With the possibility of artificial intelligence being integrated into every aspect of our life it is possible for humans to become less humane due to the abundance of technology and constantly being surrounded by it. Economic issues could arise because our society becomes increasingly dependent on electronics and metals. This would force countries to reevaluate their trade and structure more of their resources around building their technology.
Steve Omhundro explains the problem with highly intelligent AI. Picture a chess robot which is considered a rational agent (has well defined goals and takes steps to achieve that goal) (Omhundro). Of course the goal of the robot is to beat other opponents but suppose its creators decide that it’s time to cut the robot off. That interferes with the robot’s goals so it will “generate sub goals to keep itself from being unplugged” (Omohundro). Eventually if they continue to try to turn the robot off the robot will try to eliminate the threat. However this advancement in technology will bring along with is more productivity and higher efficiency.
Artificial intelligence pays for itself over time and is more cost efficient that a human worker. Robots allow humans to not put themselves in dangerous situations. The robot can simply be told to do an activity that has the potential to be dangerous and thus prevents human death. However, with the rapid growth of artificial intelligence there is a chance it could evolve. I am arguing the highly controversial topic of, with expeditious and immense growth of artificial intelligence there is a possibility that it will surpass humans in intelligence. Computer scientists such as John Searle and Alan Turing argue against strong AI.
Strong artificial intelligence is the view that suitable programmed computers can understand language and possess the same mental capabilities as humans (Stanford). Weak artificial intelligence is the view that computers are only useful in some areas because they can mimic human mental abilities (Stanford). In 1980 John Searle published “The Chinese Room Argument” to prove that artificial intelligence appears to understand language but it actually does not understand. The argument is set in a scenario in which a computer follows a program written in the computing language.
A human types Chinese symbols but does not actually understand Chinese and because the computer does what the human does it does not show understanding of Chinese either. The Turing test was created in 1950 by Alan Turing to deal with the question can machines think. It is also known as the Imitation Game and is comprised of a person, machine and interrogator. The interrogator is in a separate room from the person and the machine and the purpose of the game is for the interrogator to determine which one is the person and which is the machine.
The person and the machine are labeled X and Y and the interrogator must ask them questions and the machine is trying to make the interrogator think that it is the person. “I believe that in about fifty years’ time it will be possible to program computers, with a storage capacity of about 109, to make them play the imitation game so well that an average interrogator will not have more than 70 percent chance of making the right identification after five minutes of questioning.
I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted” said Turing (Stanford). The test may not be good because it only bases intelligence off of being able to have a conversation. However I do concede to the fact that systems are able to appear that they are intelligent. Take the program WATSON for instance. WATSON was the famous robot that beat the world’s best Jeopardy player, Ken Jennings.
WATSON gave the illusion that it was intelligent but in reality it was just faster at searching the internet for answers than Ken could provide them. Although this does happen quite a bit, it is still possible for artificial intelligence to possess some type of actual intelligence and eventually become smarter than humans. In respect to the Turing Test I can understand that it is difficult for programs to have the same mental capabilities and respond exactly like humans but they can learn from us. Just like chatboxes learn from humans artificial intelligence with more computing power can mimic our actions and learn from us and understand.
Both Turing’s and Searle’s arguments are set in a limited context. Now let us examine the systems reply to Searle’s Chinese argument. The systems reply has two parts, it concedes to the fact that the man in the room does not understand Chinese, but the man is merely a part of larger system that contains a database and instructions for understanding Chinese (Stanford). To put things simply the system actually does understand Chinese. Kurzweil says that “Searle is contradicting himself in saying in effect, “the machine speaks Chinese but doesn’t understand Chinese.
Searle responded by saying the man can attempt to remember the system, go somewhere and return but still won’t understand the Chinese or attach meaning to the symbols. However, there is an alternative to this, the human could actually memorize the book. Imagine two books one called Type 1 and Type 2. Type 2 is pretty much a Chinese-English manual so the person could learn and understand some Chinese. Of course the Type 2 would have pictures and meanings for the human to associate the Chinese symbols with so that renders Searle’s argument ineffective.
This means that strong Artificial Intelligence is indeed plausible. The Turing Test is considered successful if a computer is mistaken as a human for more than thirty percent of the conversation. On June, 7 2014 a computer program named Eugene Goostman that represents a thirteen year old Ukranian Boy convinced thirty-three percent of the judges that it was a human. This is the first time a machine has passed the Turing Test. Artificial Intelligence is well on its way to becoming more intelligent than humans. Even Alan Turing himself admitted that a program would pass his test eventually.
So now it is evident that artificial intelligence actually could one day surpass us human intelligence you are probably wondering what’s going to happen once we are no longer at the top of the hierarchy. Whether it be humanoids walking among us or giant supercomputers controlling the entire manufacturing industry we must find a way to coexist peacefully. With the evolution of their intelligence comes a conscience and the ability to think for themselves. Obviously we cannot enslave them or we run the risk of having a repetition of Nat Turner’s rebellion.
One possible solution is that we grant them the same rights as humans and possibly having restrictions on what artificial intelligence is able to do. Of course there are some of you that oppose the idea of allowing them to roam among us freely because they could be dangerous and the idea of a real life Matrix is horrifying. While there are those that do agree that the evolution of AI has the potential to be dangerous there could be fail safes integrated into the programming of every form of artificial intelligence.
At the point when artificial intelligence begins to manifest into a superior form programmers can hardwire certain rules such as they can’t harm humans or steal. Going back to the exponential growth graph representing the growth of AI, the likelihood of a Singularity is highly possible. With every breakthrough made technology continues to advance and grow faster. Looking at our society, our technology has grown immensely within the past fifty years and continues to without any sign of slowing down. Those that believe the Singularity will happen agree that strong AI will come as a result of it.
Not only do they believe in that but individuals like Ray Kurzweil welcome the fact that this is coming and aren’t trying to find ways to hinder it. Thinking about how life may be in the distant or near future can be intimidating but I encourage you to embrace it. The idea of having something else as or more intelligent than humans is completely possible and relevant. The University of Maryland has created a robot named Baxter that is able to learn how to do tasks, such as cooking, from YouTube videos. In previous experiments the robot has been able learn how to manipulate objects through observation.
It accomplishes this by a Convolutional Neural Network which enables it to break down the actions in a video and mimic those actions. With this type of technology why wouldn’t artificial intelligence be able to surpass us in intelligence? As time continues to pass and more breakthroughs in computer science occurs it will become more evident that I and a plethora of others are correct; that with the expeditious and immense growth of artificial intelligence there is a possibility that it will surpass humans in intelligence. Then the question would not be can AI become smarter than us, but how will we as a human race proceed.
As Ray Kurzweil said “There will be no distinction, post-Singularity, between human and machine or between physical and virtual reality. ” There is no doubt that humans will have to in some way establish dominance over strong AI because if not what will stop them for evolving and eventually asserting their dominance. This is quite the conundrum, there is obviously a huge risk associated with this process of artificial intelligence become stronger and that risk is what is preventing some from embracing the idea that a new era based around AI will form. All the human race can do at this point is prepare for the inevitable.