Greek version
The article below is written, not for the danger of intelligent machines attempting to obliterate humanity, but for the way realer danger of a massive rise of unemployment following the production of the first intelligent robots. Nothing would prevent the states and corporations from ignoring the existence of humans and stopping economically supporting the unemployed ones. After all, humanism is nothing more than an investment. The research must stop as soon as possible.
This is one of my first articles written in English. It's not my mother tongue so, please, excuse me for any syntax or grammar mistakes, or improper use of words and expressions.
---------------------------------------------------------------------
Introduction
Generally speaking, scientific research has positive results. Even though the progress in science has rather harmed some professions (and, according to many, the environment), its' impact on our way of life and society is advantageous. More or less, all of the everyday appliances of the 21st century owe their existence to the discovery of a group of physicists or chemists, and we owe our easy, compared to previous eras, living, almost exclusively to them.
One can find hundreds of scientific theories and achievements involved in everyday-use technologies. When it comes to the internet, for instance, (which's story begins when some scientists of CERN developed a protocol that would enable them to access the data of a particular computer of their network using the others), where wireless technology and satellite communication is used, one can say that it owes its' advance to a large degree to the comprehension of electromagnetism via Maxwell's theory.
One other example is the processors of our computers. These processors are the "brains" of our PCs, consisting of transistors instead of neurons, and they wouldn't exist if there weren't scientists experimenting with the doping of Germanium and Pyridium to create the semi-conductors with which the diods that form the transistors would be created. Without progress on semi-conductor physics and chemistry involved in doping, you wouldn't be online now.
One can find numerous examples of technological applications based on scientific discoveries. Our society owes much to scientists, indeed. But everything on this world has its' negative aspect along with the positive one. In particular, many of the scientific discoveries to which we owe our everyday appliances have been exploited for uses of doubtful morality.
The infrared radiation detectors may be used for detecting surviving people under ruins after an earthquake, but they are also being used for the detection of bombing targets, which in many cases are mere civilians. The progress of nuclear physics may have led to the -much more efficient and friendlier to the environment- use of nuclear power for the generation of electricity, but it also led to nuclear weapons.
Many scientists out there lack morals and humanism. When, during the 1st world war, some German and British scientists were requested to develop chemical weapons, no moral, not even the fear that those weapons could be found and be used by the enemy against their own children, prevented them from recommending the use of chlorine, yperite and phosgene against the enemy.
Humans have proved that, not only they are not mature enough to use, but they may not even deserve scientific advancements at all. Nevertheless, generally speaking, the scientific progress has improved the quality of our lives. Nowadays, however, it has been reached a point where one can ask... until when will science be improving our lives? Could we have reached a point to some scientific fields, where further research will do more harm than good? Let me elaborate.
When one talks about "science" doesn't necessarily mean physics and chemistry. When most people hear about "scientists", they imagine glasses-wearing people working with test tubes or solving abstract mathematics. There are branches that don't involve test tubes. However, what can one call a scientific field, depends on what one considers the definition of science to be. One can claim that a lawyer is a scientist as well, since he has to keep in his brain much knowledge related to his field and is supposed to solve problems. My "definition" for science, is whatever field includes experiments as well as theoretical knowledge.
But no matter what one calls a science, all fields related to the understanding of the human brain belong to them for sure. One extend of those fields is the one related to artificial intelligence and neural networks. And the research being done in that field is intense, for many problems demand the existence of machines combining the intelligence of a human with the computing power of a computer in order to be solved. One has to be able to discern the difference between computing power and intelligence, and to grasp what neural networks are and what makes their development so difficult and potentially useful, to be able to understand the dangers of such a scientific breakthrough. Let's have a brief description of what they are and the difficulties of developing them.
Computers and intelligence
These are the questions we will be concerned with in this part the article. We are not going to write about them as much as we should -this is not the purpose of this publication- but they are worth replying. As we previously mentioned, one has to understand some basic principles of this field in order to understand the dangers of artificial intelligence.
The question regarding the definition of intelligence is a rather philosophical one (as philosophical as "what is life?"), so we will reply the others first. A rough definition of intelligence will be given after the readers will have understood the difference between human cognition and computing power. Let's start by examining how does a computer "understand" something (whether a file is legal or not, for instance) and "think" or take a decision, and what happens in our brains before they carry out the same tasks.
Is a computer "intelligent"? This question sounds childish. Whoever is aware of some basic principles of informatics, may laugh at someone asking this question in earnest. Not everyone is aware of how computers work, though; furthermore, many people don't really know what being "intelligent" is and what's its' difference from being computationally precise. This kind of ignorance often leads people to have several absurd beliefs.
Many people out there believe that the computing systems belonging to large corporations (google, for instance), "spy" on them. These systems collect and process a vast amount of personal information, indeed, but that is not exactly "spying". A computer scanning your e-mails is way different than a real human reading them, for the simple reason that a computer does not understand what is your e-mail about. Let me elaborate.
When one sends an e-mail to somebody, there is a large possibility (since most e-mail services do so) that the e-mail will be scanned by the e-mail service's computers. By the word "scanned" we definitely don't mean that some employee of the service is going to read your e-mails and write down his conclusions about you, before giving them to some security agency. Though there is some information about such "scanning" taking place frequently and without a search warrant, it has absolutely nothing to do with what scanning is.
When a computer scans an e-mail it, in the first place, looks for defined keywords. There are algorithms seeking for particular number-series consisting of "0" and "1" (all data in computers and on the internet is series of zeros and ones) that correspond to particular words. Such keywords are
useful for the e-mail service, for all the ads it shows you have to be based on your interests, and they can only find them via your e-mails.
These keywords are predefined in the algorithm. One of them could be, for instance, the word "shoes". When the e-mail service's systems find out that you've sent an e-mail containing one of the keywords, the ads that it'll be showing you will probably regard them; the algorithm that detected the word saves in a database that you are interested in "shoes" and another algorithm uses this information to deliver to you ads related to the keyword. That results to you clicking on them -given you are indeed interested- and the e-mail service making profit.
Do you understand the difference between this procedure and a real person reading your e-mails?
The algorithms used do not understand that you are interested in shoes. They are not being led to conclusions about yourself given what you've written. They are mere detectors of particular series of 0s and 1s. They don't think something about you. They don't know what the other algorithms do. They don't comprehend the usefulness of detecting the keywords. They don't know what you've written in the e-mail and what you've been discussing with the receiver.
Let's see one more example of how computers "spy" on your e-mails.
Computers are able to detect images, besides keywords. Actually, every type of file; videos, music, text, every form of data is detectable by a computing system. Have you ever wondered how does youtube find out you uploaded a copyrighted video or that one of your videos contains a copyright-protected file? How does it know which is it and whom it belongs to? Such detections are possible thanks to some powerful, intricate mathematical algorithms concerned with assigning to each file a unique number. These mathematical algorithms are called "hash functions" and are an advanced topic in computer science.
A hash function is a mathematical algorithm that uses series of numbers ("0" and "1" in information technology) to create another, that one cannot use to find out the initial digit-series, no matter how well does one know the steps followed by the algorithm. You give in a number-series. You take out another. It's technically infeasible to use the last one to find out the one you gave. Period. How is this possible? Let the mathematicians know.
A common application in which hush functions are being used is the saving of passwords in databases. When you create an account on a social network and choose a password, it's not the password itself that is being saved; it's its' hash (the proper term is "message digest", but "hash" is also widely used) -the series of digits that the hash function produces when one is given. Whenever you type your password on a social network it gets hashed down to the series of digits the function gives out, and the result is compared to the message digest saved in the database.
An amazing aspect of those algorithms is that the output numbers -the hashes or message digests- are absolutely unique for each input. There are no multiple inputs leading to the same message digest. Hash algorithms are developed in ways that there is no possibility for such collisions to happen, so all the problems that they'd cause will be avoided. It is impossible to log in typing a wrong password that'd lead to the message digest of yours.
Just like a series of 0s and 1s corresponding to a password can be hashed to a message digest, all files can have a unique identification number generated by a hash algorithm. The e-mail services' computing systems hash all images attached to an e-mail and seek for their message digests in a database to find out information regarding them. That's how pornography is being found out by the services.
But there is a problem: each and every pixel of an image corresponds to a series of 0s and 1s. Therefore one can easily change its' hash by editing some of its' pixels. To prevent pornographers from slightly editing the pictures before they upload them to get around getting caught, it's not the pictures themselves that are being hashed, but their very structures.
Consider the image below
If we hash it out and find out its' message digest we will get an utterly unique number corresponding to it. We will be finding it out when it's uploaded by using it, ever after. But if someone edits out several of its' pixels, we will get a different hash than the original. What should we do to ensure we will detect the image even after editing?
The main method i'm aware of is resizing the picture to a size of 8x8 pixels:
and then edit the colors out:
and that's the image we are going to hash. As you see, the last image is merely the very structure of the original one. The same structure would we get if we edited parameters of our first image. Keep in mind that that's one of the methods in use. There are others I haven't even heard of and definitely more effective. This particular is a little bit simple-minded; if we resize the image in a way that the face will change position in it, it won't be detected by its' hash number for its' structure will have been changed, and therefore the message digest will be different. When we google out on similar images there are several algorithms working all together. But what counts here is understanding how systems detect images in general.
Still not understood the difference than paying a human to censor e-mails?
The computing machines do not comprehend what the image depicts. They do not know what is it about -they just search for their hash in a database, and if they find it out attached with the tag "pornography", they disallow you from sending the e-mail. They do not "see" the image. They are not aware of the e-mail's discussion topic. They only see numbers.
The ways of detecting copyrighted video or music files are alike: finding out the very structure using similar methods and hashing it off.
Here becomes more apparent how different having humans spying on you is, from having computers hashing off your images. Computers are, as one of my teachers used to say, unbelievably idiotic machines, capable only of adding and rolling binary numbers -and all the results they produce are merely specific series of these numbers in specific positions.
Humans have a way different method of perceiving images. Humans do not have to hash an image off to a message digest and find out its' tag in a database to understand what is it about. They see the images directly.
Consider the image above. It shows a face. The faces we see in our everyday life do not become message digests saved in our brains, but are remembered in such a way that we are able to recognize them even when we see them from different angles or edited -something that computers are incapable of doing. If you saw the profile of the face showed on the image above you would realise that it's the same person, even though without having ever seen it.
The same would happen if we cropped and pasted a part of the face in a random image; the background would differ but your brain would still recognize it's a familiar face, although you haven't seen it in front of such a background. How is this possible? Does out brain "hash" the face in order to detect it when it's seen, just like a computer does with an image? How come and we recall it even when we see it with another expression or under other circumstances?
The science concerned with such questions is known as "cognitive science". Cognitive science is a combination of several scientific fields regarding human brain function and cognition. These sciences include neurology, psychology, psychiatry, anthropology, linguistics, neurobiology and even computer science. The latter has helped us realise what an intricate system our thinking machine is, since whenever one tries to make a computer mimic our brain's capabilities, one realises that functions seemingly so simple that no one is impressed of them (i.e. recognizing an image) are proved to be extremely tough for a machine to implement.
When we view an image, our brain forms a "concept" about it. Whatever are elementary particles for physics and atoms for chemistry, the same are concepts for cognitive science. Everything stored in our brain is stored as a concept. Technically, all information we have in our heads is chemical substances between interconnected neurons that form patterns depending on the knowledge being absorbed by our brains, and are activated whenever we seek for that stored information. But that's what happens in neurochemical level; in less fundamental levels one can talk about "concepts".
So, whenever one sees a face, a concept is created in his brain. This concept is in all cases connected with other ones, less or more related to it -they could be names, memories, emotions etc. No single concept is unconnected with the others. So, when a concept about a face is being created it's also connected with other, related ones.
A face can have a particular characteristic -blue eyes for instance. In such a case, the concept created for that face is connected with the concept "blue eyes". That's how you are able to remember someone when seeing a random person, if their faces share a common characteristic. Your first love could have an attribute like blue eyes, and that may have resulted to you connecting it with an emotion. That emotion may be somehow activated when you see a person with that characteristic.
When we say than a concept is activated, we mean that the neural interconnections containing it are receiving electrochemical impulses transferred through the neural axons of other brain cells. Well, it's easier to talk about concept activations. These activations are responsible for your reasoning.
A concept created for a particular face does not regard only characteristics such as the eye color. It also contains information about its' symmetry and its' precise mathematical analogies (i.e. length of eyes compared to length of upper part of nose). That's why if we edited such parameters of a familiar face you'd realise something is wrong.
The reason of the difficulty of the development of artificial neuronal networks is exactly this -the concept formation. How can one make a machine create a concept given an amount of data? Furthermore, how can you make a machine able to conceptualize, capable of reasoning logically? How can you make it understand what an e-mail is about instead of just detecting some words?
The information saved in an average human brain is so much that it is more possible that seeing something would lead to utterly illogical thoughts, instead of reasonable ones. One could see a face with blue eyes with that resulting of him remembering a blue-colored painting, which would remind him paints and that, in turn, would lead to him thinking of... paintbrushes. Are there many normal persons thinking of paintbrushes when they see a face? Though it may happen to some eccentric artists, one can't help but ask how come and we don't think of absolutely irrelevant things when seeing something since our head contains more concepts irrelevant to it, rather than relevant ones.
The answer is apparently that concepts do not work as simply as one may have thought; in particular, it's obvious that the relevance of a concept to another is of great significance. By technical aspect, it depends on how strong the neural interconnections are between the two parts of a brain in which those concepts are saved. Given this, how can we make a computer capable of logically reasoning? How can we make it understand whether two concepts have anything to do with each other? How can we persuade it that a the concept A is more relevant to the concept X rather than Y?
Well, we could achieve it by somehow making their physical interconnection stronger than it is. But this is a task our brain does automatically -our brain somehow understands which concepts are more relevant to whom, an obvious sign of intelligence. Would we be working with an "intelligent" machine if we had to strengthen its' conceptual interconnections manually?
How can we make a computer capable of detecting images by understanding what they depict, instead of hashing them off and search for tags attached to their message digests in a database? How can we make it understand what the interests of one are, by comprehending one's e-mails instead of just detecting series of "0" and "1"?
As you may have realised, these problems require the in-depth understanding of our brain function in order to be solved, so that's (thankfully) not something to be achieved in the foreseeable future. But computer science has helped scientists to make some progress on that topic; the attempts to solve seemingly trivial problems (i.e. translations using computers) have made them realise our brain does not work as simply as they once believed. And several computer-related applications can help one make some comparisons.
We mentioned previously the function of hash-algorithms and the editing of images in a way that only their very structures would remain, before being hashed off to a message digest. Could our brain follow some similar steps before storing a concept in our memory? "Similar", meaning that much information (like a file consisted of a large series of 0s and 1s) gets hashed off -reduced- in a unique way to something simpler (just like a message digest), that can be recognized even after being changed.
Likely. After all, if that didn't happen we wouldn't be able to recognize familiar images or concepts when seen in another form. We wouldn't recognize a melody we've heard if played with another instrument or in a different pace, for instance, nor would we recognize a face if we'd see it with another expression. We can't help but boil down to the conclusion that whatever is stored in our brain, is so in such a simplified form that it's possible for us to recognize it no matter the circumstances and the slight changes to it -the only requirement is to be the same with what it is stored. Just like computers are able to recognize an image even after it's edited, unless it's edited in a way that it's structure is changed, and therefore it's not the same image.
Will computers ever be able to conceptualize information as we do? Will they ever be capable of performing the same tasks with our brains? Will they ever understand what we tell them when we speak, and what are we talking about in an e-mail? All of these will be feasible if the research on artificial intelligence continues for some decades. And they will all be huge scientific achievements, indeed.
Cognitive science is one the most interesting fields of science. This section's purpose was to give you an idea of what makes the development of intelligent machines so difficult and how different is our way of perception than a computer's (if we can even call it so). We may have not yet answered all of the questions we should, but after reading this your brain may have "formed a concept" of the answers -even if it's not explicit ;)
I'll try to give a definition of intelligence some paragraphs below. What we should keep in mind at this point is that computers are better than the human brain, only in terms of speed and precision. They are incapable of thinking and understanding, and they their performance in all tasks is inferior to the one of our minds. Fortunately, the human cognition is so complicated that no machine will beat it in the near future.
Intelligent machines
The idea of intelligent-robots -robots with the intelligence of a human and the computing power of a computer- excites those who are not aware of their dangers. Everyone has heard about the progress in the field of artificial intelligence and most people consider talking robots to be part of our everyday life in the future. Some people, however, claim that the development of intelligent machines/robots will be the beginning of the end of the domination of humans on earth. Some of them think that those machines will even attempt to obliterate us. How many movies with such plot have you watched?
As we are going to see, even though such an outcome is possible, the end of our domination on earth (or at least of our societies) will not come that way. Though there is a possibility that paranoid scenarios of wrongly programmed robots turning to psychopaths that enjoy killing humans or of the exploitation of their technology by persons wanting to become global dictators, will come true, the main danger of developing such technology is the complete replacement of human workers by intelligent robots -a scenario worse than anyone can even dare to imagine.
What will the uses of artificial brains be? What's the point of developing intelligent machines, anyway, and why is so much money being spent to the research on it? Aren't the computers we use enough for our tasks and the problems one has to cope with?
Well, as we've previously seen, the modern computers are of limited capabilities; they are unbelievably idiotic machines able only to add and roll numbers -and particularly binary numbers, not even the ones we use. Our computers have great computing power in terms of speed and precision, but do not even show the slightest sign of intelligence and creativity, and that means that all problems are still being solved by humans that just exploit computers. How would it be if they were instantly solved by computers?
For various reasons, some people are determined to ensure that in the future the machines will be carrying out all labour being done nowadays by humans. It's a pity that there are many advocates of turning our world to such a place. Their dream is to slack off all day long, having robot-servants doing everything for them -a disgusting dream that seems more like a nightmare to me. And humanity will have to pay a high price if this dream ever come true.
The intelligent robots -all artificial neural networks- will not only be used as slaves or servants; their existence requires the development of an artificial brain mimicking human cognition. That means that every problem, regardless of its' nature, will be solved by them, for they'll have the computing capabilities of a computer along with our intelligence. So, we would use such machines for every application that requires human labour.
One I can think of (at least for the first neural networks in work) is the fixing of bugs in source codes of computer applications -for not all of them are found by human programmers. The more extensive and carefully written a code is, the more difficult to find its' bugs and back-doors. Doing so for requires not only intelligence, but also superhuman memory capabilities. When it comes to software consisted of thousands of command lines, there work groups of programmers, having all of the problems of working with a group.
Wouldn't it be better of we could give the software to an intelligent machine, having our cognition and a computer's computing capabilities, to fix all errors automatically? Even better, wouldn't it be more practical to have intelligent machines developing all those codes (before being capable to run them by imagining them)? As years will be passing by, all of this labour will be carried out by intelligent machines, given that they'll become at some point equally intelligent with us. But what is it going to happen to the jobless programmers, in particular?
Whatever will have happened to all unskilled workers -they'll all have become unemployed a long time before programmers get replaced by robots. Neural networks will be capable of developing software several years after they'll be capable of doing all jobs that don't require particular skills. We will see robots working as cashiers sooner than we'll see them working as programmers. What is it going to happen to the unskilled ones, then?
And let's say that the first years having intelligent robots to work for you will worth a fortune. But thanks to the progress on the field, it will sooner or later become significantly cheaper than having human workers. What is going to happen then?
The replaced unskilled workers will be useless to the corporation owners and our states. There will be nothing for them to do, for all of what they were doing will be carried out now by robots. What will prevent then our states from ignoring their existence? What will be the point of economically supporting them? After all, there is the huge problem of overpopulation waiting to be solved -a problem that nobody dares to talk about. That's because the only ways to solve it include, if not a war, definitely restrictions on fundamental human rights (i.e. the right to have as many kids as you want and when you want them). But it's a very real problem, and soon it'll have to be solved. Closing your eyes to avoid seeing the danger does not protect you from it. What if they choose to solve it by stopping supporting the unemployed ones and letting them die starving?
Perhaps the scenario described in the above paragraph sounds absurd, a paranoid conspiracy theory. It isn't. On the contrary, it is a very possible outcome, given the earnings they'll have. It's been repeatedly proven in history that whenever one has all of one's need satisfied, one does not show great sensibility when seeing the others struggling. Another proof of how insensible they can become are is their indifference while demanding from states to fire their public sector employees even though they don't have the slightest chance of getting another job, and to ruin their enterprises, all of this for the sake of the so-called economic recession. What makes you believe that they won't be proved to be even more misanthrope then? After all, no one is going to stop them from programming their robots to kill whoever is going to try to do something against their actions. Yeah, the scenario I previously rejected as paranoid seems plausible now.
An intelligent robot will be a more efficient worker than a human; it will carry out all labour better, it'll have no demands due to its' lack of biological needs (unless it has to eat to keep its' neurons active, if its' intelligence is based on a neural network), it'll be tireless and never ask for anything. Having such workers available will fundamentally change our society.
One could claim that the above paragraphs are wrong for the simple reason that with most ex-workers -most humans- penniless, there would be no point of keeping industry and economy alive. The corporation-owners and the states would stop having earnings. Therefore, they would still give money to the unemployed ones. Though this argument is reasonable and logical, whoever might use it wouldn't keep in mind the reason for which the ones using robot-employees are in need of money and earnings. The reason is their basic needs and whatever needless they consider to be necessary. Well, these needs would be satisfied without us buying their products. Money would be useless and the economy would cease to exist. What's the point of the earnings when you have robot-servants and the ability to have whatever you want for free?
They wouldn't do something for you or me. After all, to them, humanism is an investment aiming to ensure the constant availability of numerous and therefore inexpensive employable workers. That would be ensured with the production of robots, and therefore humanism would be pointless.
As time will be passing by, and due to the poorness of most people and the ability of the rich to have all of their needs fulfilled without human workers and even money, the economy would stop having any reason to exist and would fall apart. It sounds absurd, but it's the most possible outcome if robots replace human workers.
To me, such a society would be disgusting, for people would have become lazy, crap-producing machines slacking around all day long. People must have a purpose of existence; this cannot only be the satisfaction of their biological needs, but also doing something useful for society via their labour. If all problems were to be solved by machines... what would our life be like? Can you even imagine being constantly laid off and having others doing all you were supposed to? But let's just ignore the moral part of this and consider such a society to be a "wonderful" place to live...
Consider what intelligence is. Everything you read in the previous part about computers and brains. We can define intelligence as the ability of a system -either neuronal, like our brains, or consisting of transistors- to form concepts when given information and use them to reason. Given this, think of an intelligent being with the computing power of a super-computer.
One so highly intelligent life-form wouldn't only have the ability to multiply quickly; it would be able to think just like you and me. It would have a personality, an ability to think about itself and its' interests, and to make plans for its' future. Technically, such a machine would have a soul, since what we call a "soul" is merely the result of our brain activity. Our thoughts, our behaviour, our reactions... everything is the result of the electrochemical activity of our brains, and when is stops, what we call our soul ceases to exist. And those robots would have a soul similar to ours.
What makes you believe that a machine with a soul would agree with being our obedient servant? Expecting intelligent creatures that would be even more intelligent than we are, and wouldn't have a need to eat (maybe besides some pills to feed their neurons), to be our slaves is... illogical. You can have a human being carrying out your labour if you help it to satisfy its' biological need to feed itself by paying it. How can you persuade a machine to be your servant since it has no needs? After all, it would have more important things to do; it would be aware its' cognitive superiority and not only it would not allow to serve less-intelligent creatures (would you accept being the servant of a chimp?), but it would be trying to achieve... whatever would an intelligent being want to achieve. Understanding the universe or... dominating the planet, for instance.
I am not saying that they would try to obliterate us. It is indeed a possible scenario, but its' possibility is as high as accepting to become our servants: low. For the same reason they wouldn't be our slaves -it would be pointless! They would have no reason to do it, since they don't have needs like ours. All wars in history started either due to economical and social reasons (technically, for our biological needs) either to religious. Such machines would have no biological needs and would be intelligent. Therefore it is more likely that they would not start a war against us -unless they'd have needs like ours or some fundamental error in the development of their brains would turn them to psychopaths. But even if they wouldn't start a war, what would prevent them from forming their own societies?
It may sound absurd, but it's likely they would choose to do so; just like we, humans, formed societies consisted of members of our species -similar to us when it comes to needs and behaviour (talking, thinking etc)-, robots might prefer forming societies consisting of members of their kind.
So, we wouldn't have robot-slaves and they would not bother serving us. That's partly because, as we previously mentioned, there would be no way to reward them; you reward a human being by paying it in order to have its' needs satisfied -that's how it's motivated to carry out labour. How could one apply the same principle to robots? They would not bother serving or killing us and they would probably prefer forming their own societies. But that would apparently cause some rivalry between them and us.
At this point, it seems more plausible that we'd start a war; if humans hate each other for reasons such as their national pride, religion, the social classes to which they belong (though this particular is not that absurd), everyday life issues or even football teams, how would they feel and react if intelligent machines started forming their own societies? A rivalry would be inevitable. But if there started a war... well, nothing would prevent the intelligent machines from producing more of themselves and even dominating our planet.
Some of you might argue that it would be pointless for those machines to form a society within ours for the simple reason that they would have no biological needs unable to fulfil by themselves -the main reason for which humans became "social" animals- if they even had. On the contrary, the only thing that would give a meaning to their existence would be serving us, humans. That's true indeed, and the possibility of them distinguishing themselves from us is limited. But the very structure of their brains (or whatever would they have instead of one) could be such that they would tend to be as social as we are, if not more. There is also some possibility that the existence of robots with no needs like ours will be infeasible. Therefore, it's a plausible scenario that they'd form their own societies or even try to obliterate us, if they started considering us to be enemies as we tend to do with other human beings.
But even if they accepted to be our servants and there would be no problems deriving from the unemployment of millions of humans, the result of having them to do everything would be the genetic and mental degeneration of humans.
We've already mentioned what will our society turn into if robots replace human workers and the states keep helping the unemployed ones. In such a case, everyone will just live a life slacking around and have all of his labour carried out by robots. Though some may argue that in such a society we would have all time we needed to do whatever we'd consider creative, well, it isn't so. Do you really believe that the vast majority of people have really creative things to do? Writing novels or symphonies, for instance, or seeking out the meaning of life and its' definition? Just consider how do most spend their leisure time. Even the creative ones would prefer doing nothing; too much slacking around turns humans into animals. Being laid off for some periods of your life is good and enjoyable, but being constantly so is maddening.
Such a life-style would degenerate humans both morally and mentally -they would have machines thinking for them and therefore no reason to use their own brains (and in such a disgusting society those machines would literally censor your e-mails for they would then understand what are they about). But besides the purely moral part of the story -what does a human whose only interest is producing feces worth?- there is also the serious problem of our biological degeneration. What we call "natural selection" is not something taking place in wild-life only; all species tend to adjust themselves to the conditions under which they live -technically changing their genetic make-up to live easily. Our genetic evolution still goes on in our society. What genetic-garbage would be created several thousands of millions of years after humans have machines doing everything for them, having no need even to think?
Is that what we want for our kind? Degeneration? Keep in mind that that's the good scenario; it's more possible that after humans would become useless to the rich ones, their population would be dramatically reduced, for reasons we elaborated. The outcome then would be the degeneration of those remaining.
Conclusions
The development of intelligent machines will be a colossal scientific achievement. Such a break-through will be even more admirable than the understanding of human brain, which is, after all, a requirement of developing artificial brains. But such a breakthrough will be detrimental to our society and species; it will be proved to be a Pandora's box that one way or another, it will ruin our kind.
Maybe the scenarios described above sound like paranoid conspiracy theories. They are not; the dangers of developing such technology are unfortunately very real, and the more the research goes on, the realer they become. It would be exceptionally interesting to develop an artificial brain purely for research purposes, but humanity is not mature enough to use such creations. The development of such technology will be the beginning of the end for humanity, for humans will not use it for research purposes only, but exploit it for economic ones.
The scientific progress is, generally speaking, advantageous to our way of life. But that is not true when it comes to intelligent machines; the ruining of every structure of society as we know it would be inevitable. The cons of developing artificial brains are way more than the possible pros, and they are merely the results of their initially seeming advantageous social impact, which will be proved short-lived.
The research on artificial intelligence must stop. It's as simple as that. The scientists working on that field must realise how many the dangers are and stop working on their projects, as soon as possible.
Maybe in the distant future, in an era that people will be aware of the impact of their actions and will care more about the collective interest of humanity in the long-term, rather than their own in the short-term, a continuance of the research will not be so ruinous.
---------------------------------------------------------------------
©George Malandrakis
All rights reserved
The article below is written, not for the danger of intelligent machines attempting to obliterate humanity, but for the way realer danger of a massive rise of unemployment following the production of the first intelligent robots. Nothing would prevent the states and corporations from ignoring the existence of humans and stopping economically supporting the unemployed ones. After all, humanism is nothing more than an investment. The research must stop as soon as possible.
This is one of my first articles written in English. It's not my mother tongue so, please, excuse me for any syntax or grammar mistakes, or improper use of words and expressions.
---------------------------------------------------------------------
Introduction
Generally speaking, scientific research has positive results. Even though the progress in science has rather harmed some professions (and, according to many, the environment), its' impact on our way of life and society is advantageous. More or less, all of the everyday appliances of the 21st century owe their existence to the discovery of a group of physicists or chemists, and we owe our easy, compared to previous eras, living, almost exclusively to them.
One can find hundreds of scientific theories and achievements involved in everyday-use technologies. When it comes to the internet, for instance, (which's story begins when some scientists of CERN developed a protocol that would enable them to access the data of a particular computer of their network using the others), where wireless technology and satellite communication is used, one can say that it owes its' advance to a large degree to the comprehension of electromagnetism via Maxwell's theory.
How much do we owe to the
comprehension of electromagnetism?
(image from blog.agupieware.com)
One can find numerous examples of technological applications based on scientific discoveries. Our society owes much to scientists, indeed. But everything on this world has its' negative aspect along with the positive one. In particular, many of the scientific discoveries to which we owe our everyday appliances have been exploited for uses of doubtful morality.
The infrared radiation detectors may be used for detecting surviving people under ruins after an earthquake, but they are also being used for the detection of bombing targets, which in many cases are mere civilians. The progress of nuclear physics may have led to the -much more efficient and friendlier to the environment- use of nuclear power for the generation of electricity, but it also led to nuclear weapons.
Many scientists out there lack morals and humanism. When, during the 1st world war, some German and British scientists were requested to develop chemical weapons, no moral, not even the fear that those weapons could be found and be used by the enemy against their own children, prevented them from recommending the use of chlorine, yperite and phosgene against the enemy.
A scientific discovery may have
ruinous results too.
(image from genius.com)
Humans have proved that, not only they are not mature enough to use, but they may not even deserve scientific advancements at all. Nevertheless, generally speaking, the scientific progress has improved the quality of our lives. Nowadays, however, it has been reached a point where one can ask... until when will science be improving our lives? Could we have reached a point to some scientific fields, where further research will do more harm than good? Let me elaborate.
When one talks about "science" doesn't necessarily mean physics and chemistry. When most people hear about "scientists", they imagine glasses-wearing people working with test tubes or solving abstract mathematics. There are branches that don't involve test tubes. However, what can one call a scientific field, depends on what one considers the definition of science to be. One can claim that a lawyer is a scientist as well, since he has to keep in his brain much knowledge related to his field and is supposed to solve problems. My "definition" for science, is whatever field includes experiments as well as theoretical knowledge.
But no matter what one calls a science, all fields related to the understanding of the human brain belong to them for sure. One extend of those fields is the one related to artificial intelligence and neural networks. And the research being done in that field is intense, for many problems demand the existence of machines combining the intelligence of a human with the computing power of a computer in order to be solved. One has to be able to discern the difference between computing power and intelligence, and to grasp what neural networks are and what makes their development so difficult and potentially useful, to be able to understand the dangers of such a scientific breakthrough. Let's have a brief description of what they are and the difficulties of developing them.
Computers and intelligence
- What is intelligence?
- Why is a computer not intelligent?
- What makes the development of artificial neural networks and brains so difficult?
- Why are computers incapable of mimicking human cognition? What makes the human brain function so special?
- Will machines ever be able to think, feel, reason logically, understand and translate from one language to another?
These are the questions we will be concerned with in this part the article. We are not going to write about them as much as we should -this is not the purpose of this publication- but they are worth replying. As we previously mentioned, one has to understand some basic principles of this field in order to understand the dangers of artificial intelligence.
The question regarding the definition of intelligence is a rather philosophical one (as philosophical as "what is life?"), so we will reply the others first. A rough definition of intelligence will be given after the readers will have understood the difference between human cognition and computing power. Let's start by examining how does a computer "understand" something (whether a file is legal or not, for instance) and "think" or take a decision, and what happens in our brains before they carry out the same tasks.
Is a computer "intelligent"? This question sounds childish. Whoever is aware of some basic principles of informatics, may laugh at someone asking this question in earnest. Not everyone is aware of how computers work, though; furthermore, many people don't really know what being "intelligent" is and what's its' difference from being computationally precise. This kind of ignorance often leads people to have several absurd beliefs.
Many people out there believe that the computing systems belonging to large corporations (google, for instance), "spy" on them. These systems collect and process a vast amount of personal information, indeed, but that is not exactly "spying". A computer scanning your e-mails is way different than a real human reading them, for the simple reason that a computer does not understand what is your e-mail about. Let me elaborate.
Being spied by electronic systems is
(for the time being) way different than this.
(image found on techiecode.wordpress.com)
When a computer scans an e-mail it, in the first place, looks for defined keywords. There are algorithms seeking for particular number-series consisting of "0" and "1" (all data in computers and on the internet is series of zeros and ones) that correspond to particular words. Such keywords are
useful for the e-mail service, for all the ads it shows you have to be based on your interests, and they can only find them via your e-mails.
These keywords are predefined in the algorithm. One of them could be, for instance, the word "shoes". When the e-mail service's systems find out that you've sent an e-mail containing one of the keywords, the ads that it'll be showing you will probably regard them; the algorithm that detected the word saves in a database that you are interested in "shoes" and another algorithm uses this information to deliver to you ads related to the keyword. That results to you clicking on them -given you are indeed interested- and the e-mail service making profit.
Do you understand the difference between this procedure and a real person reading your e-mails?
The algorithms used do not understand that you are interested in shoes. They are not being led to conclusions about yourself given what you've written. They are mere detectors of particular series of 0s and 1s. They don't think something about you. They don't know what the other algorithms do. They don't comprehend the usefulness of detecting the keywords. They don't know what you've written in the e-mail and what you've been discussing with the receiver.
Let's see one more example of how computers "spy" on your e-mails.
Computers are able to detect images, besides keywords. Actually, every type of file; videos, music, text, every form of data is detectable by a computing system. Have you ever wondered how does youtube find out you uploaded a copyrighted video or that one of your videos contains a copyright-protected file? How does it know which is it and whom it belongs to? Such detections are possible thanks to some powerful, intricate mathematical algorithms concerned with assigning to each file a unique number. These mathematical algorithms are called "hash functions" and are an advanced topic in computer science.
A hash function is a mathematical algorithm that uses series of numbers ("0" and "1" in information technology) to create another, that one cannot use to find out the initial digit-series, no matter how well does one know the steps followed by the algorithm. You give in a number-series. You take out another. It's technically infeasible to use the last one to find out the one you gave. Period. How is this possible? Let the mathematicians know.
A common application in which hush functions are being used is the saving of passwords in databases. When you create an account on a social network and choose a password, it's not the password itself that is being saved; it's its' hash (the proper term is "message digest", but "hash" is also widely used) -the series of digits that the hash function produces when one is given. Whenever you type your password on a social network it gets hashed down to the series of digits the function gives out, and the result is compared to the message digest saved in the database.
An amazing aspect of those algorithms is that the output numbers -the hashes or message digests- are absolutely unique for each input. There are no multiple inputs leading to the same message digest. Hash algorithms are developed in ways that there is no possibility for such collisions to happen, so all the problems that they'd cause will be avoided. It is impossible to log in typing a wrong password that'd lead to the message digest of yours.
Just like a series of 0s and 1s corresponding to a password can be hashed to a message digest, all files can have a unique identification number generated by a hash algorithm. The e-mail services' computing systems hash all images attached to an e-mail and seek for their message digests in a database to find out information regarding them. That's how pornography is being found out by the services.
But there is a problem: each and every pixel of an image corresponds to a series of 0s and 1s. Therefore one can easily change its' hash by editing some of its' pixels. To prevent pornographers from slightly editing the pictures before they upload them to get around getting caught, it's not the pictures themselves that are being hashed, but their very structures.
Consider the image below
(image: http://www.hackerfactor.com/)
If we hash it out and find out its' message digest we will get an utterly unique number corresponding to it. We will be finding it out when it's uploaded by using it, ever after. But if someone edits out several of its' pixels, we will get a different hash than the original. What should we do to ensure we will detect the image even after editing?
The main method i'm aware of is resizing the picture to a size of 8x8 pixels:
and then edit the colors out:
and that's the image we are going to hash. As you see, the last image is merely the very structure of the original one. The same structure would we get if we edited parameters of our first image. Keep in mind that that's one of the methods in use. There are others I haven't even heard of and definitely more effective. This particular is a little bit simple-minded; if we resize the image in a way that the face will change position in it, it won't be detected by its' hash number for its' structure will have been changed, and therefore the message digest will be different. When we google out on similar images there are several algorithms working all together. But what counts here is understanding how systems detect images in general.
Still not understood the difference than paying a human to censor e-mails?
The computing machines do not comprehend what the image depicts. They do not know what is it about -they just search for their hash in a database, and if they find it out attached with the tag "pornography", they disallow you from sending the e-mail. They do not "see" the image. They are not aware of the e-mail's discussion topic. They only see numbers.
The ways of detecting copyrighted video or music files are alike: finding out the very structure using similar methods and hashing it off.
Here becomes more apparent how different having humans spying on you is, from having computers hashing off your images. Computers are, as one of my teachers used to say, unbelievably idiotic machines, capable only of adding and rolling binary numbers -and all the results they produce are merely specific series of these numbers in specific positions.
Humans have a way different method of perceiving images. Humans do not have to hash an image off to a message digest and find out its' tag in a database to understand what is it about. They see the images directly.
(image from www.goldenmeancalipers.com)
The same would happen if we cropped and pasted a part of the face in a random image; the background would differ but your brain would still recognize it's a familiar face, although you haven't seen it in front of such a background. How is this possible? Does out brain "hash" the face in order to detect it when it's seen, just like a computer does with an image? How come and we recall it even when we see it with another expression or under other circumstances?
The science concerned with such questions is known as "cognitive science". Cognitive science is a combination of several scientific fields regarding human brain function and cognition. These sciences include neurology, psychology, psychiatry, anthropology, linguistics, neurobiology and even computer science. The latter has helped us realise what an intricate system our thinking machine is, since whenever one tries to make a computer mimic our brain's capabilities, one realises that functions seemingly so simple that no one is impressed of them (i.e. recognizing an image) are proved to be extremely tough for a machine to implement.
When we view an image, our brain forms a "concept" about it. Whatever are elementary particles for physics and atoms for chemistry, the same are concepts for cognitive science. Everything stored in our brain is stored as a concept. Technically, all information we have in our heads is chemical substances between interconnected neurons that form patterns depending on the knowledge being absorbed by our brains, and are activated whenever we seek for that stored information. But that's what happens in neurochemical level; in less fundamental levels one can talk about "concepts".
So, whenever one sees a face, a concept is created in his brain. This concept is in all cases connected with other ones, less or more related to it -they could be names, memories, emotions etc. No single concept is unconnected with the others. So, when a concept about a face is being created it's also connected with other, related ones.
A face can have a particular characteristic -blue eyes for instance. In such a case, the concept created for that face is connected with the concept "blue eyes". That's how you are able to remember someone when seeing a random person, if their faces share a common characteristic. Your first love could have an attribute like blue eyes, and that may have resulted to you connecting it with an emotion. That emotion may be somehow activated when you see a person with that characteristic.
When we say than a concept is activated, we mean that the neural interconnections containing it are receiving electrochemical impulses transferred through the neural axons of other brain cells. Well, it's easier to talk about concept activations. These activations are responsible for your reasoning.
A concept created for a particular face does not regard only characteristics such as the eye color. It also contains information about its' symmetry and its' precise mathematical analogies (i.e. length of eyes compared to length of upper part of nose). That's why if we edited such parameters of a familiar face you'd realise something is wrong.
The reason of the difficulty of the development of artificial neuronal networks is exactly this -the concept formation. How can one make a machine create a concept given an amount of data? Furthermore, how can you make a machine able to conceptualize, capable of reasoning logically? How can you make it understand what an e-mail is about instead of just detecting some words?
The information saved in an average human brain is so much that it is more possible that seeing something would lead to utterly illogical thoughts, instead of reasonable ones. One could see a face with blue eyes with that resulting of him remembering a blue-colored painting, which would remind him paints and that, in turn, would lead to him thinking of... paintbrushes. Are there many normal persons thinking of paintbrushes when they see a face? Though it may happen to some eccentric artists, one can't help but ask how come and we don't think of absolutely irrelevant things when seeing something since our head contains more concepts irrelevant to it, rather than relevant ones.
The answer is apparently that concepts do not work as simply as one may have thought; in particular, it's obvious that the relevance of a concept to another is of great significance. By technical aspect, it depends on how strong the neural interconnections are between the two parts of a brain in which those concepts are saved. Given this, how can we make a computer capable of logically reasoning? How can we make it understand whether two concepts have anything to do with each other? How can we persuade it that a the concept A is more relevant to the concept X rather than Y?
Well, we could achieve it by somehow making their physical interconnection stronger than it is. But this is a task our brain does automatically -our brain somehow understands which concepts are more relevant to whom, an obvious sign of intelligence. Would we be working with an "intelligent" machine if we had to strengthen its' conceptual interconnections manually?
An artistic depiction of a neural network.
No machine is capable of mimicking
even the simplest functions of one.
(image from www.pageresource.com)
As you may have realised, these problems require the in-depth understanding of our brain function in order to be solved, so that's (thankfully) not something to be achieved in the foreseeable future. But computer science has helped scientists to make some progress on that topic; the attempts to solve seemingly trivial problems (i.e. translations using computers) have made them realise our brain does not work as simply as they once believed. And several computer-related applications can help one make some comparisons.
We mentioned previously the function of hash-algorithms and the editing of images in a way that only their very structures would remain, before being hashed off to a message digest. Could our brain follow some similar steps before storing a concept in our memory? "Similar", meaning that much information (like a file consisted of a large series of 0s and 1s) gets hashed off -reduced- in a unique way to something simpler (just like a message digest), that can be recognized even after being changed.
Likely. After all, if that didn't happen we wouldn't be able to recognize familiar images or concepts when seen in another form. We wouldn't recognize a melody we've heard if played with another instrument or in a different pace, for instance, nor would we recognize a face if we'd see it with another expression. We can't help but boil down to the conclusion that whatever is stored in our brain, is so in such a simplified form that it's possible for us to recognize it no matter the circumstances and the slight changes to it -the only requirement is to be the same with what it is stored. Just like computers are able to recognize an image even after it's edited, unless it's edited in a way that it's structure is changed, and therefore it's not the same image.
Will computers ever be able to conceptualize information as we do? Will they ever be capable of performing the same tasks with our brains? Will they ever understand what we tell them when we speak, and what are we talking about in an e-mail? All of these will be feasible if the research on artificial intelligence continues for some decades. And they will all be huge scientific achievements, indeed.
Cognitive science is one the most interesting fields of science. This section's purpose was to give you an idea of what makes the development of intelligent machines so difficult and how different is our way of perception than a computer's (if we can even call it so). We may have not yet answered all of the questions we should, but after reading this your brain may have "formed a concept" of the answers -even if it's not explicit ;)
I'll try to give a definition of intelligence some paragraphs below. What we should keep in mind at this point is that computers are better than the human brain, only in terms of speed and precision. They are incapable of thinking and understanding, and they their performance in all tasks is inferior to the one of our minds. Fortunately, the human cognition is so complicated that no machine will beat it in the near future.
Intelligent machines
- How great is the possibility of them becoming a threat to humanity?
- How will they affect our lives and what will their impact on our societies be?
- What is going to happen to the millions of unemployed when intelligent machines will have their jobs taken?
- Will humans be enslaved by robots?
The idea of intelligent-robots -robots with the intelligence of a human and the computing power of a computer- excites those who are not aware of their dangers. Everyone has heard about the progress in the field of artificial intelligence and most people consider talking robots to be part of our everyday life in the future. Some people, however, claim that the development of intelligent machines/robots will be the beginning of the end of the domination of humans on earth. Some of them think that those machines will even attempt to obliterate us. How many movies with such plot have you watched?
As we are going to see, even though such an outcome is possible, the end of our domination on earth (or at least of our societies) will not come that way. Though there is a possibility that paranoid scenarios of wrongly programmed robots turning to psychopaths that enjoy killing humans or of the exploitation of their technology by persons wanting to become global dictators, will come true, the main danger of developing such technology is the complete replacement of human workers by intelligent robots -a scenario worse than anyone can even dare to imagine.
Robots attempting to obliterate humans? Though it is a possible scenario,
the end of the society as we know it will have different causes.
(image found on www.destructoid.com)
What will the uses of artificial brains be? What's the point of developing intelligent machines, anyway, and why is so much money being spent to the research on it? Aren't the computers we use enough for our tasks and the problems one has to cope with?
Well, as we've previously seen, the modern computers are of limited capabilities; they are unbelievably idiotic machines able only to add and roll numbers -and particularly binary numbers, not even the ones we use. Our computers have great computing power in terms of speed and precision, but do not even show the slightest sign of intelligence and creativity, and that means that all problems are still being solved by humans that just exploit computers. How would it be if they were instantly solved by computers?
For various reasons, some people are determined to ensure that in the future the machines will be carrying out all labour being done nowadays by humans. It's a pity that there are many advocates of turning our world to such a place. Their dream is to slack off all day long, having robot-servants doing everything for them -a disgusting dream that seems more like a nightmare to me. And humanity will have to pay a high price if this dream ever come true.
The intelligent robots -all artificial neural networks- will not only be used as slaves or servants; their existence requires the development of an artificial brain mimicking human cognition. That means that every problem, regardless of its' nature, will be solved by them, for they'll have the computing capabilities of a computer along with our intelligence. So, we would use such machines for every application that requires human labour.
One I can think of (at least for the first neural networks in work) is the fixing of bugs in source codes of computer applications -for not all of them are found by human programmers. The more extensive and carefully written a code is, the more difficult to find its' bugs and back-doors. Doing so for requires not only intelligence, but also superhuman memory capabilities. When it comes to software consisted of thousands of command lines, there work groups of programmers, having all of the problems of working with a group.
Wouldn't it be better of we could give the software to an intelligent machine, having our cognition and a computer's computing capabilities, to fix all errors automatically? Even better, wouldn't it be more practical to have intelligent machines developing all those codes (before being capable to run them by imagining them)? As years will be passing by, all of this labour will be carried out by intelligent machines, given that they'll become at some point equally intelligent with us. But what is it going to happen to the jobless programmers, in particular?
Whatever will have happened to all unskilled workers -they'll all have become unemployed a long time before programmers get replaced by robots. Neural networks will be capable of developing software several years after they'll be capable of doing all jobs that don't require particular skills. We will see robots working as cashiers sooner than we'll see them working as programmers. What is it going to happen to the unskilled ones, then?
And let's say that the first years having intelligent robots to work for you will worth a fortune. But thanks to the progress on the field, it will sooner or later become significantly cheaper than having human workers. What is going to happen then?
The replaced unskilled workers will be useless to the corporation owners and our states. There will be nothing for them to do, for all of what they were doing will be carried out now by robots. What will prevent then our states from ignoring their existence? What will be the point of economically supporting them? After all, there is the huge problem of overpopulation waiting to be solved -a problem that nobody dares to talk about. That's because the only ways to solve it include, if not a war, definitely restrictions on fundamental human rights (i.e. the right to have as many kids as you want and when you want them). But it's a very real problem, and soon it'll have to be solved. Closing your eyes to avoid seeing the danger does not protect you from it. What if they choose to solve it by stopping supporting the unemployed ones and letting them die starving?
Perhaps the scenario described in the above paragraph sounds absurd, a paranoid conspiracy theory. It isn't. On the contrary, it is a very possible outcome, given the earnings they'll have. It's been repeatedly proven in history that whenever one has all of one's need satisfied, one does not show great sensibility when seeing the others struggling. Another proof of how insensible they can become are is their indifference while demanding from states to fire their public sector employees even though they don't have the slightest chance of getting another job, and to ruin their enterprises, all of this for the sake of the so-called economic recession. What makes you believe that they won't be proved to be even more misanthrope then? After all, no one is going to stop them from programming their robots to kill whoever is going to try to do something against their actions. Yeah, the scenario I previously rejected as paranoid seems plausible now.
A robot-servant! Isn't it a great idea!? It won't seem
so great when such machines will have your jobs taken.
(image from sharewhale.com)
An intelligent robot will be a more efficient worker than a human; it will carry out all labour better, it'll have no demands due to its' lack of biological needs (unless it has to eat to keep its' neurons active, if its' intelligence is based on a neural network), it'll be tireless and never ask for anything. Having such workers available will fundamentally change our society.
One could claim that the above paragraphs are wrong for the simple reason that with most ex-workers -most humans- penniless, there would be no point of keeping industry and economy alive. The corporation-owners and the states would stop having earnings. Therefore, they would still give money to the unemployed ones. Though this argument is reasonable and logical, whoever might use it wouldn't keep in mind the reason for which the ones using robot-employees are in need of money and earnings. The reason is their basic needs and whatever needless they consider to be necessary. Well, these needs would be satisfied without us buying their products. Money would be useless and the economy would cease to exist. What's the point of the earnings when you have robot-servants and the ability to have whatever you want for free?
They wouldn't do something for you or me. After all, to them, humanism is an investment aiming to ensure the constant availability of numerous and therefore inexpensive employable workers. That would be ensured with the production of robots, and therefore humanism would be pointless.
As time will be passing by, and due to the poorness of most people and the ability of the rich to have all of their needs fulfilled without human workers and even money, the economy would stop having any reason to exist and would fall apart. It sounds absurd, but it's the most possible outcome if robots replace human workers.
But let's say that producing intelligent robots wouldn't have such an outcome. Let's say that all labour will be carried out by them, that we wouldn't have to cope with the previously mentioned problems and that our societies would turn to wonderful places in which we would all be concerned with everything but our so-tiring professions, and that we would be getting everything for free thanks to the availability of robots producing everything tangible.
Could robots be proved to be friends?
Maybe. But not without consequences.
(image from www.boymeetsrobots.com)
Consider what intelligence is. Everything you read in the previous part about computers and brains. We can define intelligence as the ability of a system -either neuronal, like our brains, or consisting of transistors- to form concepts when given information and use them to reason. Given this, think of an intelligent being with the computing power of a super-computer.
One so highly intelligent life-form wouldn't only have the ability to multiply quickly; it would be able to think just like you and me. It would have a personality, an ability to think about itself and its' interests, and to make plans for its' future. Technically, such a machine would have a soul, since what we call a "soul" is merely the result of our brain activity. Our thoughts, our behaviour, our reactions... everything is the result of the electrochemical activity of our brains, and when is stops, what we call our soul ceases to exist. And those robots would have a soul similar to ours.
What makes you believe that a machine with a soul would agree with being our obedient servant? Expecting intelligent creatures that would be even more intelligent than we are, and wouldn't have a need to eat (maybe besides some pills to feed their neurons), to be our slaves is... illogical. You can have a human being carrying out your labour if you help it to satisfy its' biological need to feed itself by paying it. How can you persuade a machine to be your servant since it has no needs? After all, it would have more important things to do; it would be aware its' cognitive superiority and not only it would not allow to serve less-intelligent creatures (would you accept being the servant of a chimp?), but it would be trying to achieve... whatever would an intelligent being want to achieve. Understanding the universe or... dominating the planet, for instance.
I am not saying that they would try to obliterate us. It is indeed a possible scenario, but its' possibility is as high as accepting to become our servants: low. For the same reason they wouldn't be our slaves -it would be pointless! They would have no reason to do it, since they don't have needs like ours. All wars in history started either due to economical and social reasons (technically, for our biological needs) either to religious. Such machines would have no biological needs and would be intelligent. Therefore it is more likely that they would not start a war against us -unless they'd have needs like ours or some fundamental error in the development of their brains would turn them to psychopaths. But even if they wouldn't start a war, what would prevent them from forming their own societies?
It may sound absurd, but it's likely they would choose to do so; just like we, humans, formed societies consisted of members of our species -similar to us when it comes to needs and behaviour (talking, thinking etc)-, robots might prefer forming societies consisting of members of their kind.
What will prevent intelligent robots from forming their own societies or
even starting a war against us? Due to their nature, the possibility of
such an outcome is low. But still existent.
(image from www.livescience.com)
So, we wouldn't have robot-slaves and they would not bother serving us. That's partly because, as we previously mentioned, there would be no way to reward them; you reward a human being by paying it in order to have its' needs satisfied -that's how it's motivated to carry out labour. How could one apply the same principle to robots? They would not bother serving or killing us and they would probably prefer forming their own societies. But that would apparently cause some rivalry between them and us.
At this point, it seems more plausible that we'd start a war; if humans hate each other for reasons such as their national pride, religion, the social classes to which they belong (though this particular is not that absurd), everyday life issues or even football teams, how would they feel and react if intelligent machines started forming their own societies? A rivalry would be inevitable. But if there started a war... well, nothing would prevent the intelligent machines from producing more of themselves and even dominating our planet.
Some of you might argue that it would be pointless for those machines to form a society within ours for the simple reason that they would have no biological needs unable to fulfil by themselves -the main reason for which humans became "social" animals- if they even had. On the contrary, the only thing that would give a meaning to their existence would be serving us, humans. That's true indeed, and the possibility of them distinguishing themselves from us is limited. But the very structure of their brains (or whatever would they have instead of one) could be such that they would tend to be as social as we are, if not more. There is also some possibility that the existence of robots with no needs like ours will be infeasible. Therefore, it's a plausible scenario that they'd form their own societies or even try to obliterate us, if they started considering us to be enemies as we tend to do with other human beings.
But even if they accepted to be our servants and there would be no problems deriving from the unemployment of millions of humans, the result of having them to do everything would be the genetic and mental degeneration of humans.
We've already mentioned what will our society turn into if robots replace human workers and the states keep helping the unemployed ones. In such a case, everyone will just live a life slacking around and have all of his labour carried out by robots. Though some may argue that in such a society we would have all time we needed to do whatever we'd consider creative, well, it isn't so. Do you really believe that the vast majority of people have really creative things to do? Writing novels or symphonies, for instance, or seeking out the meaning of life and its' definition? Just consider how do most spend their leisure time. Even the creative ones would prefer doing nothing; too much slacking around turns humans into animals. Being laid off for some periods of your life is good and enjoyable, but being constantly so is maddening.
Such a life-style would degenerate humans both morally and mentally -they would have machines thinking for them and therefore no reason to use their own brains (and in such a disgusting society those machines would literally censor your e-mails for they would then understand what are they about). But besides the purely moral part of the story -what does a human whose only interest is producing feces worth?- there is also the serious problem of our biological degeneration. What we call "natural selection" is not something taking place in wild-life only; all species tend to adjust themselves to the conditions under which they live -technically changing their genetic make-up to live easily. Our genetic evolution still goes on in our society. What genetic-garbage would be created several thousands of millions of years after humans have machines doing everything for them, having no need even to think?
Is that what we want for our kind? Degeneration? Keep in mind that that's the good scenario; it's more possible that after humans would become useless to the rich ones, their population would be dramatically reduced, for reasons we elaborated. The outcome then would be the degeneration of those remaining.
Conclusions
The development of intelligent machines will be a colossal scientific achievement. Such a break-through will be even more admirable than the understanding of human brain, which is, after all, a requirement of developing artificial brains. But such a breakthrough will be detrimental to our society and species; it will be proved to be a Pandora's box that one way or another, it will ruin our kind.
Maybe the scenarios described above sound like paranoid conspiracy theories. They are not; the dangers of developing such technology are unfortunately very real, and the more the research goes on, the realer they become. It would be exceptionally interesting to develop an artificial brain purely for research purposes, but humanity is not mature enough to use such creations. The development of such technology will be the beginning of the end for humanity, for humans will not use it for research purposes only, but exploit it for economic ones.
The scientific progress is, generally speaking, advantageous to our way of life. But that is not true when it comes to intelligent machines; the ruining of every structure of society as we know it would be inevitable. The cons of developing artificial brains are way more than the possible pros, and they are merely the results of their initially seeming advantageous social impact, which will be proved short-lived.
The research on artificial intelligence must stop. It's as simple as that. The scientists working on that field must realise how many the dangers are and stop working on their projects, as soon as possible.
Maybe in the distant future, in an era that people will be aware of the impact of their actions and will care more about the collective interest of humanity in the long-term, rather than their own in the short-term, a continuance of the research will not be so ruinous.
---------------------------------------------------------------------
All rights reserved
Δεν υπάρχουν σχόλια:
Δημοσίευση σχολίου