When I was working on comics I had a side novel that I was working on. Now that I am working full time on the novels, this one is going to be pushed back a little bit. This was my first attempt at “non-fiction”.
I figured I’d put it up here in case I never get back to it.
Chapter X – Rise of the Machines
I’ve never been comfortable with the idea of intelligent machines or ‘artificial intelligence’. Perhaps it was all the science fiction movies of the 80’s like Terminator and even Aliens with its artificial human that taught me to be prejudiced against thinking robots.
What made the original Matrix movie so interesting was that it took the idea of conqueror robots to the next level; not only did the conquer us, they enslaved us and turned us into an energy source. The world of comic-books is a little more diverse in it’s depiction of conscious computers. At least the creators of those colourful stories were able to show not only the villainous side of human-hating robots but also the value and emotional turmoil of a robot with a consciousness. Characters such as Marvel’s Vision and DC’s Red Tornado would often put their artificial lives on the line for humanity. Yet it is the cold mechanical efficiency of the T-800 and Agent Smith that tends to remain in our consciousness even after the good guys, humanity, have won.
The fear of our creations coming to life and killing everyone has always existed in our myths and folklore. The first story that comes to mind in our modern mythology would be the Golem of Prague. In the early versions of the story the Golem would continually repeat simple tasks which would then result in tragedy, such as filling a house with water when no one ordered it to stop. As the story evolved with each retelling the Golem became lethal, eventually evolving into a murderous monster. One of the more entertaining story arcs that stretched through multiple seasons of Star Trek: The Next Generation was that of Data and his brother Lore. Data, the benevolent member of Starfleet and Lore, the mischievous and dangerous brother, were often at odds with each other. Both were created by the same scientist but Lore had something that Data would not receive until the series was cancelled; an emotion chip that would allow him to experience real human feelings. What made their conflict so entertaining was that Lore had feelings whereas Data did not. Yet Data was somehow driven by a greater good within him to stop his brother’s evil plots. So the question is not whether or not machines will become intelligent because inevitably they will be able to at the very least match human intelligence. First we have to acknowledge the difference between an intelligent machine and a sentient being. The question becomes will they develop emotion, compassion and feeling and if they do will they be benevolent?
BLOOD AND METAL
As of January 2012, robots are being used to kill humans. Known as Unmanned Aerial Vehicles or “drones”, these computer guided machines have been used to strike terrorist targets in Pakistan, primarily high ranking Taliban officials. Some claim these drones are more accurate and therefore reduce the amount of collateral damage, effectively making war less dangerous for bystanders. Others like Lord Bingham, former Law Lord of Britain, compared these drone strikes to landmines and cluster bombs which are no longer used on the modern battlefield for humane reasons. (1) However these computerized predators are not intelligent machines, not in the sense that they are fully autonomous and making their own decisions. Someone still has to be the one to hit the “kill switch”, the decision of whether or not to drop the bomb and obliterate the target is still being made by a human being. This could change at any time because the technology is available, it’s simply a question of when those in control of the military robots decides a human is no longer needed.
Even if the military robots that are being built over the next decade are capable of autonomous military operations, without an artificial intelligence there is no real murder or justice. A machine is built to follow a set of instructions programmed into it by humans. Even if you program a robot to kill it will have no more knowledge of what it has done than a bulldozer would that crushed a person. Without having knowledge of self, you cannot have empathy for others and would not understand what it has done. A machine that has been programmed to create a situation that leads to the death of a human is no more responsible than a person who ties a string to a gun trigger and aims the gun at the door for the next person who enters. For a machine to actively murder humans it must be conscious of itself and conscious of the fact it is ending a human life. This is something the robots of today are not capable of doing.
Science Fiction writer Vernor Vinge came up with a theory known as the technological singularity. Basically, it hypothesizes that at some point in the future technology will reach a point where it will evolve into a super-intelligence far greater than that of any human (2). A.I. is one of the more popular reasons for this event to occur, although there are others. Vernor Vinge and fellow writer Ray Kurtzweil believe it is hard for us to predict or understand exactly when and how this event might take place. They believe the super-intelligence that arises from this event will be beyond our comprehension, in other words they believe the “AI moves in mysterious ways”, that what we have created will quickly surpass even our imaginations. Most supporters of the theory believe the Singularity will happen this century using the rapid growth of computer power in the last fifty years as proof of an upward trend.
“Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an ‘intelligence explosion,’ and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make.” – I.J. Good (5)
One theory given as to how this super-intelligent computer will come to exist is through self-improvement. The theory goes that eventually we will build a machine that can improve itself, either by updating its existing programming or by replicating itself into a newer, more intelligent machine. This theory raises an interesting question; if two machines are created and each are identical except for a .01 percent variation in their programming, with each duplication and upgrade there is the possibility of chaotic behaviour affecting the outcome. In other words if an intelligent computer begins evolving, eventually the machines will begin to differentiate from each other. Self-awareness might not come by way of self-improvement at all, but through random mutation. If a Singularity Event was to occur, even if the machines were to become sentient, there is a very real chance they might not all think alike. Humanity would find itself suddenly surrounded by thinking machines that no longer wanted to work together and could no longer communicate and identify with each other. For all their intelligence they would get along about as well as the rest of us do.
There are two institutions investigating the benefits of friendly A.I. They are the Singularity Institute for Artificial Intelligence and the Future of Humanity Institute. Billions of dollars are being spent every year on A.I. research but few involved are willing to accept the notion that intelligent machines may not be perfect and that intelligence itself does not make one superior. Whether or not a Singularity will occur is still up for debate. Perhaps humanity will somehow manage to knock itself backwards in technology with a devastating war or massive disaster strikes. Intelligent machines are coming whether we like it or not, but what about sentient machines? At what point do they become a threat to us? Or will they ever?
THE FIFTH GENERATION
The Fifth Generation Computer Systems project (FGCS) was initiated by Japan’s Ministry of International Trade and Industry in 1982 (3). It’s mission was to create a “fifth generation computer”, one that would usher in a new era of intelligent computers. The Minister was determined to put Japan at the forefront of computer. Japan had already conquered the home electronics market in the 70’s and it looked as if they were about to do the same in the automotive industry and confidence in the project was high. One of the goals of this project was to create platform upon which to build artificial intelligence. After ten years and $400 million dollars U.S. the project was deemed a failure. The greatest Japanese minds with government funding and global support were unable to create anything even closely resembling a “Fifth Generation” of computers. There are so many challenges involved with creating smart computers and even though the technology has improved exponentially since the 80’s, the exact same challenges continue to rear their heads. There was one issue that Japanese programmers had to tackle and that was the language problem, something the Americans did not have to face. There are only twenty six letters in the English alphabet and each character can be encoded into one byte. In the Japanese language there are over 256 characters. The Japanese instead use a “double byte” or “multi-byte” encoding system. This could explain why we did not see a massive leap forward in technology in the Japanese computer industry similar to what we saw in consumer electronics. In the film 2001: A Space Odyssey a computer could speak and interact directly with humans by the year 1992. Twenty years later the machines are still mostly silent. They may not have made the leaps in computer technology that they had wanted but they created a generation of talent that would continue working on similar projects well into the 21st century.
In robotics Japan has once again taken the lead, spending more money annually than any other country. You can find entertainment robots like ASIMO, built by Honda or QRIO which was built by Sony. There are androids, actroids and robot dogs like AIBO. The Japanese Science and Technology Agency has created what has been dubbed the creepiest robot ever made, CB2 (4). CB2 is a result of the current trend towards building robots along more natural lines. CB2 was built to imitate the behaviour and appearance of a human child of about two years old. It has a Biomemetic Body that allows it to feel through special sensors in its “skin”. It was built by a frightening combination of robotic engineers, brain specialists and psychologists. What makes CB2 a breakthrough in robotic technology is its ability to remember and imitate facial expressions. If the scientists had thought this would make it more human, they did not take into account the Uncanny Valley hypothesis that states that when human replicas act like, but not identical too, real human beings, it causes of sense of revulsion in those that encounter it. This uncanny valley theory has been used to explain everything from our cultures current fascination with zombies to the failure of the movie The Polar Express. Sigmund Freud mentioned something called the Uncanny in a 1919 essay, saying that something that is familiar can be alien at the same time. Perhaps it is simply a natural reaction to something which appears to be a lifeless, dead version of ourselves. http://en.wikipedia.org/wiki/File:Mori_Uncanny_Valley.svg
That which is pleasing to the eye survives, while that which we deem undesirable is inevitably destroyed. SKYNET had it wrong in the Terminator movies. A giant robot skeleton looks to viewers like a Ray Harryhausen production. The truly scary robots are the creepy little ones made to look like children!
Researchers at MIT have created a robot called KISMET, what they call a sociable humanoid robot (7). The purpose of KISMET and others like it is to interact directly with humans. Build upon familiar models of child development, KISMET is designed to be sociable, intuitive and can mimic the behaviour of its teachers. One of the features of KISMET is the ability to perceive personal space, when a subject moves to close it appears to withdrawal with its expression. Moving away or moving too fast will make KISMET display an irritated face. When someone speaks to it, it will raise its eyebrows as if listening intently. In the future robots might learn to use these simple tricks to appear more human or perhaps they might have a genuine interest in what we have to say after all.
The Uncanny Valley theory is going be play a huge part in any future relations between mankind and intelligent robots. Researchers in A.I. originally believed they could create an intelligent computer simply by inputting all the knowledge they could store into a machine that would potentially mimic the human brain. But the brain is more complex than they thought. The subject of the human brain is something that will be touched on later in the book. Instead research has focused on creating a more naturally evolving intelligence, one that grows and learns. Living creatures are born with some basic instinct on how to exist. Machines do not have that gift. A baby is born with an attachment to its mother and will bond with whoever gives it sustenance. A machine cannot tell what it is looking at when giving cameras to peer through; it has no concept of anything other than what parameters are programmed into it.
THE THIN LINE
“The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else.” -Eliezer Yudkowsky (6)
There are probably thousands of books written about creating a smarter machine. Research and development into artificial emotion and self-awareness are fairly new concepts in the world of robotics research. Emotion is necessary for the development of any kind of decision-making ability, we simply cannot function without a clear sense of what is important. Existence it seems is all about prioritizing. Feelings of hunger drive you to a food supply, fear leads to flight or fight responses, exhaustion inevitably leads to sleep. These innate naturally evolved processes that we take for granted cannot be duplicated. An artificial intelligence cannot develop without some sense of self-preservation and emotion. The machines from Terminator were ordered by Skynet to wipe out humanity but Skynet only lashed out when it felt that it would be terminated. David the robot boy from Stanley Kubrick’s film A.I. was intelligent and emotional but the film revolved around his need to understand how his ‘mother’ could abandon him. David believed if he could just see his mother again she would want to keep him. David seemed to need human companionship even in a world populated by other thinking, feeling robots. We assume our human brains come equipped with a well defined ability to make sense of reality but this is not exactly true. As I will discuss later there are already signs of mental afflictions in machines, not counting the millions of viruses and glitches that can cause all kinds of strange behaviour in computers.
Humanity has a particular knack for humanizing objects and giving weight to inanimate objects. Cats and dogs become members of our family; cars and toys become pieces of memories and emotions from the past or representations of certain ideals we believe in. People form attachments with their machines the minute they finish building them but will intelligent machines form the same attachment to humanity? In a society where so many people lie to each other and to themselves about their true feelings will it even matter if a robot is expressing genuine emotions or not? In this regard Hollywood had it right. In order for a machine to feel love, acceptance and compassion it will also have to be programmed with the opposite ideals; hate, rejection and cruelty. It truly is a thin line between love and hate.
An artificial intelligence that can think beyond its programmed parameters will inevitably develop some kind of mental illness.
The South Korean Ministry of Justice has become the first country to use robots in the penal system (8). These three new robot guards will patrol the prison in Pohang at night, monitoring prisoner health and watching for suspicious behaviour. South Korea is pushing ahead with groundbreaking robotics programs in order to raise global interest in their innovations. Unlike their human counterparts, these robots are designed to cause as little tension among the prisoners as possible. It is hard to feel oppressed or abused by a robot that resembles a big grinning WALL-E. These robot guards can sense hostility, injury and violence among the inmates but will not directly instigate these events. The human guards are not to blame; it is a tough, dangerous and stressful job. The three robots cost almost $900,000 U.S. to produce so for now humans are still a more affordable solution. Although I doubt too many people will be outraged by robots replacing humans at such a dangerous task, South Korea also has many other robots ready to enter the workforce. Robotic machine guns already guard the border between North and South Korea with optical sensors that can spot targets up to 3 km away. They have also pledged to build a robot theme park and have started an education plan that will include a robot in every classroom. These teaching aids, called EngKey, are primarily there to assist human teachers but they serve a secondary function as well; their toy-like design makes children more comfortable and open around them then they might be with an adult. Keeping children interested is a vital factor in their education and so far the novelty of robots has done that. Eventually robots that cannot imitate humans will lose their charm and will become commonplace.
The FA-Men restaurant in Nagoya, Japan has robot chefs ready to serve up meals to its customers , thanks to redesigned industrial robots by AISEI (9). When there are no orders to be served the robots will engage in a bit of theatrical combat, with knives used as swords and plates as shields. In Bangkok four robots serve guests in a shabu-shabu style. Guests can select instructions from touch screens at their tables and the robots will serve their every whim; from taking away dirty dishes to dancing for their entertainment. China also has a robot restaurant in Jinan, called fittingly the Dalu Robot Restaurant. Some of their robots are nothing more than wheeled tables, while others are custom designs meant to invoke a 1950’s American sci-fi nostalgia. The robots at the Dalu Robot Restaurant are really nothing more than the same style of automatons you’d see at any children’s amusement park. The difference is that now they have taken positions of employment that had previously belonged to a human being.
From the assembly line robots that have taken over the factories, to the automated check-out desks at the grocery store, technology is changing the way people do business. Just as the economy begins a slow recovery we are poised to see massive unemployment thanks to a robotics revolution. Job loss due to robotic displacement will become a key economic issue in the next ten years. How long before we start to see HUMANS FIRST signs at political rallies? The only thing that will halt a growing robot workforce is the value of the machines. If more and more people become unemployed, who will be left to purchase these robots? They can manufacture and repair themselves but they cannot buy or employ themselves. If a wealthy robot can buy and sell his fellow robot would this be the same thing in theory as slavery?
In an e-mail conversation with Paul W. Abrahams, a man whose history dates back to the beginning days of computers and whose credentials are too long to list here, unemployment was the biggest issue facing humanity in regards to artificial intelligence (10).
“I’m not too much worried about intelligent machines taking over the world, but
I’m quite worried about the difficulty that economies are having in adjusting to
a world where unemployment becomes the norm as a result of automation. I
think the answer is to have a guaranteed minimal income that one can get
without working — and without any disgrace. You can simply choose not to
work and no one will hassle you. The wealth in our society is there to pay
for it. But if you want more than the minimum, then you have to work for it.”
Is it possible that robots will force us to create a new social structure? That the automation of the workforce will actually cause a cultural shift away from capitalism into something newer and more adaptable?
2) Superintelligence. Answer to the 2009 EDGE QUESTION: “WHAT WILL CHANGE EVERYTHING?”: http://www.nickbostrom.com/views/superintelligence.pdf
3) The fifth generation: Japan’s computer challenge to the world. Edward Feigenbaum; Pamela McCorduck. http://www.atarimagazines.com/creative/v10n8/103_The_fifth_generation_Jap.php
4) CB2 Child Robot is possibly the most disturbing machine ever built
By Conrad Quilty-Harperposted June 3rd 2007 http://www.engadget.com/2007/06/03/cb2-child-robot-is-possibly-the-most-disturbing-machine-ever-bui/
5) Good, I. J. (1965), Franz L. Alt and Morris Rubinoff, ed., “Speculations Concerning the First Ultraintelligent Machine”, Advances in Computers (Academic Press) 6: 31–88, archived from the original on 2001-05-27, retrieved 2007-08-07
6) Yudkowsky, Eliezer (2008), Bostrom, Nick; Cirkovic, Milan, eds., “Artificial Intelligence as a Positive and Negative Factor in Global Risk” (Oxford University Press)
7) KISMET http://www.ai.mit.edu/projects/humanoid-robotics-group/kismet/kismet.html
8) Robots Patrol South Korean Prison http://app.yonhapnews.co.kr/YNA/Basic/article/new_search/YIBW_showSearchArticle.aspx?searchpart=article&searchtext=%ED%8F%AC%ED%95%AD&contents_id=AKR20111123195400004
10) Paul W. Abrahams, Sc.D., CCP, is the author of “TeX for the Impatient”, MIT graduate, 1956.