Customer Reviews

The Artilect War: Cosmists vs. Terrans: A Bitter Controversy Concerning Whether Humanity Should Build Godlike Massively Intelligent Machines

5 star:		(3)
4 star:		(2)
3 star:		(3)
2 star:		(2)
1 star:		(1)

 

 

If one examines the history of research into artificial intelligence (AI), one will see it to be one of periods of incredible optimism as well as periods of extreme pessimism. Funding for research into AI, both private and public, is partly responsible for this. But the researchers themselves bear a certain measure of responsibility for the wild swings that have marked the history of AI. It seems that as soon as something is invented that appears promising or "intelligent" it is shortly thereafter abandoned as being "trivial" or uninteresting. Researchers always seem willing to go along with this unfortunately, even though they have indeed made significant progress in certain areas. Once an algorithm or reasoning pattern is understood, its status as `intelligent' is taken away and it is thereafter viewed as `just another part of the programming toolbox'. There is strong evidence, coming mostly from the commercial realm, that truly intelligent machines exist and are saving and making companies hundreds of millions of dollars in their deployment in the field. This intelligence is however not noticed or recognized as such. It is viewed merely as software that is `running' on the machines, complex yes, but not really different than `ordinary' software that has been used for decades now. If this pattern continues, then no matter how intelligent machines get they will not be viewed as being so. Their human users will therefore not be intimidated by or even impressed by them. They will become accepted just like any other piece of technology, usually taken for granted, although at times becoming an annoyance due to their need for repair and adjustment (this need becoming more critical as their complexity increases).

When this book is read with this in mind its main thesis, namely that there will sometime in the relatively near future be a controversy over the building of `massively' intelligent machines, completely dissolves. The author believes that sometime in the second half of the twenty-first century, humanity will divide itself into two camps. One of these, called the `Cosmists', will advocate the building of what the author calls `artilects', which are "massively" intelligent machines. The other camp, called the `Terrans', is strongly opposed to the building of these types of machines. The tension between these two groups will become so extreme the author argues, that it will result in a full-scale war between them, resulting in the deaths of millions of people.

In the book the author details his reasons for believing that this will happen, and he even discusses his own anxieties on the possibility of massively intelligent machines. The author is a noted expert in machine intelligence, especially in the fields of evolutionary computing and evolvable hardware. Therefore when a researcher like the author makes the claims he does it motivates the reader to examine his arguments in more detail. It is apparent when reading the book that these arguments have been carefully thought out, even though at times, because of the Cosmist-Terran terminology, the reading sometimes appears sophomoric or science-fictional in quality.

The claims made in this book would have more credibility if progress in artificial intelligence could be modeled by large discrete jumps. Central to its claim is that there will arrive a time at which both `Cosmists' and `Terrans' agree that superintelligent machines can be realized or manufactured. The apprehension felt by the Terrans will motivate them to try and suppress this realization, this behavior putting them squarely against the Cosmists. This conflict will escalate into full-scale war, fought with highly advanced and destructive technology.

But progress in AI is basically a smooth function of time, and there has been progress, despite the extreme skepticism of many individuals (most of these, again, being AI researchers themselves). Like any other field, some of the ideas in AI have not been fruitful, and have fallen by the wayside. Advances in AI have been steady, and the advances, as well as its applications are rapidly accelerating. The use of intelligent machines has become routine, so routine in fact that it is not really noticed. One can expect this trend to continue, and researchers twenty years from now will no doubt think that real intelligence has not yet been achieved. The bar will then get raised again. All the while the machines are performing useful functions and will exist in complete symbiosis with the humans around them. However, there will still be anxiety about the future arrival of superintelligent machines. It may take a while, probably till the end of the twenty-first century, for this anxiety to alleviate. Historians of technology in the first year of the twenty-second century will no doubt look back at this one and be perplexed as to why AI progress was not really part of the consciousness of those who were involved in it. These same historians will also feel another emotion when they study the developments of twenty-first century technology, including artificial intelligence:

Astonishment

This book is largely a one man debate by A.I. researcher Hugo de Garis on whether we should build massively intelligent machines later in the 21st century. Hugo presents to us a new vocabulary: artilect means artificial intellect, cosmist means those people in favor of building artilects, and terrans are against building artilects. Garis, by his own admission, describes himself as a cosmist, with misgivings, as this future technology gives him pause to reflect on it's awesome power, and whether it will be safe to build them, as regards to human safety and survivability. Consider this: an advanced artilect could easily have a mind trillions of times more powerful than a human mind, thus they may consider us vermin fit to be exterminated! Garis presents many arguments from both the cosmist and terran viewpoints. He also writes that it will be almost impossible to prevent artilects from being built. Garis believes that the primary global political question of this century will be which species will be dominant, human or artilect?

Well, Garis is much too negative for me. For a more positive spin on all of this you could read THE SINGULARITY IS NEAR by Ray Kurzweil. In his book Kurzweil writes that advanced artificial intelligences could be designed to be friendly to humans (Garis disagrees) and probably will respect their creators. Between these two books I consider Kurzweil's book to be the far superior, with many references. Garis is too self-congratulatory for my taste, and I personally believe the whole cosmist-terran debate as presented by Garis to be a non-issue. Artilects will arrive among us in a gradual way as very useful devices, and eventually become conscious beings in and of themselves, and we will be glad to have them around. I do agree with Garis on one point, however, and that is that advanced A.I. and it's applications will come to dominate 21st century economics.

De Garis argues that the ability to build 'Godlike massively intelligent machines', which he believes will be achieved sometime this century, will produce a global schism between those of us pro (Cosmists) and those of us against (Terrans) the utilisation of such technology. The result will be global war between the opposing camps, with the use of the vastly advanced weaponry which will then be available leading to what he describes as 'gigadeath'. Whether such a prophesy is realistic or purely in the realms of science fiction is debatable - the problem I have is with the motives behind de Garis writing the book the first place. De Garis tries to portray himself as a Cosmist with a conscience - he is definitely in the 'pro' camp (he is a prominent figure in the field of Artificial Intelligence and currently working on building 'brain machines'), but confesses to being tormented by nightmares of what his work may result in. To appease his conscience, de Garis therefore has published this book as a warning against what his work may lead to. I find this attitude very difficult to swallow. Considering the potential consquences of his work - the end of humanity - if de Garis has even the slightest moral doubt he should postpone his work and focus on resolving the ethical implications first. On the other hand, if he truly believes in his quest - which he compares to a religious one - he should just quietly go ahead with his work. But what de Garis is effectively doing is saying "I'm going to build something that will kill your grandchilden - try and stop me!". What is allegedly an attempt to raise public awareness comes across more as an attempt to raise de Garis's own profile.

The book is not perfect. I think that in many ways De Garis glosses over complexities that might have a big impact. For example, he does not address the fact that the emergence of intelligent machines will not be the only crisis of technology that humans will be dealing with that threaten our survival: we will also have biotech, nanotech, and who knows what else. De Garis does mention these, but not in the context of how these competing issues might impact public policy and perhaps tip the balance in favor of the "terrans", as De Garis terms them.

De Garis also makes some large leaps in the book that are probably sound but that are not substantiated. For example, he mentions "asteroid-sized" artilects without explaining how or why this scale of machine would be built. Presumably he bases it on the argument that once the machines take over they will escalate their own scale - but he doesn't make this clear and the reader might be put off by descriptions of asteroid-sized machines, when in fact this is not really relevant to the main issue. The main issue is whether to allow the creation of machines that are smarter than us.

But De Garis is right that there is a fundamental issue that is independent of the details: if we can and do create such machines, we will surely lose all control over our destiny - to the extent that we have control now.

I also agree that De Garis is right that this is an emerging issue and perhaps the most important issue of the 21st century, if one expands the issue to the larger issue of oversight of technology in general.

The are many complex social issues that need to be addressed if one is to plan for how to implement oversight over technology, and De Garis does not get into this at all. A book (or many) is needed on this topic. De Garis says clearly that his goal is not to create such a plan but rather to raise awareness of the issues and stimulate debate.

I am very glad that I read the book because in spite of the leaps it is grounded I believe, although it might seem far-fetched to someone who has not thought much about these issues before. I think that De Garis' framework for thinking about the issues is effective. I strongly recommend the book.

- Cliff

Professor De Garis book, on the Artilect War, is a philosophically thought provoking work on the possible future of humanity. I found the arguments very pursuasive and very worthy of debate not only in the Business and in the Scientific Community but also in the Political field. As I am in my early forties, I have seen the rise of technology on an ever incresing scale, what I considered science fiction when I was growing up, is now scientific fact. When I was in the middle years of High School, computer labs had only just began to make an entry, and I remember thinking how wonderful technology had become when my parents brought home one of the first home gaming stations which played tennis and squash and graphically consisted of two rectangular shapes and a square ball.


How different is the world today, then the world that existed a mere thirty years ago. If Moore's Law continues as I am sure it will, we have to be aware of the potential future that we humans have, not if but when the point of singularity is reached in the next decade or two. When this occurs there will be an exponential growth in computing power as 3d computer chips take over from the 2d chips that we currently use. With this increasing computing speed and power, by the later decades of this century Man-made Artificial Intelligences may be potentially trillions of times superior to our biological brain. The controvery both for and against building these Artificial Intelligences, is great, as we may be creating our new future, moving aware from our biological system to a silicon system, or alternatively building our potential destroyers, as our biological brain would be so slow and inferior in comparison to these god-like beings, we may be viewed by the Artilects in the same way that we view an earthworm.

I implore people to read this book, to gain knowledge, it may or may not be too late to avert the potential future proposed by Professor De Garis, but we as the current most intelligent species on the planet should be aware of the potential downside to increasing computer technology, and not be excited purely because the graphics on the new PS3 or XBox allow for greater gameplay. Potentially humanitys' existence is on the line, and Professor De Garis has some very thought provoking questions and possible future scenarios that we should be aware of. We should be debating the pros and cons of building these massively intelligent machines because we owe it to our children and to generations of the future.

I highly recommend this thought provoking book.

The author argues that one day soon, humanity will divide into two warring camps: those who favor the development of superintelligent computers (which he calls Artilects--short for artificial intellects) and the humans who oppose this development. His thesis is that the opposers will so strongly fear the possibility that the artilects will exterminate humans that they will do anything to prevent their development--even waging all-out war against the humans who favor their development. He asserts that the result of this war will be nearly a total annihilation of humanity. The book takes up all manner of counterarguments and bats them down. The oddity is that the author claims to be a developer of these artilects who is genuinely conflicted about which camp he should belong in. Yet he can't bring himself to suppress the inventions he is working on that will help bring about the catastrophe he predicts. So rather than stop the work, he writes the book as a form of penance: "Stop me before my science runs amok and kills off our species--but if you don't, I'm having a wonderful time stroking my ego as one of the foremost cybernetic thinkers and doers of our time--which may well be the end-times of our species!" One argument he doesn't deal with is what happens if nobody pays much attention to the development of artilects--which, after all, won't just be sitting around philosophizing: they'll be put to work running our air traffic system, our energy grid, and all manner of other vital functions. By the time the potential enemies of the artilects take their eyes off Britney and other pop-culture distractions, it will be too late to stop the artilect supporters or the artilects themselves. At that point, we will just have to hope the artilects are sufficiently amused by our antics not to exterminate us. This, by the way, would be before they build themselves some space vehicles and depart on an exploration of the universe, where they'll find a way to tap into the power supplies of many different suns, so they won't be dependent on any one solar system for their survival--thus achieving immortality.

The only real problem with this book is that it may have been written decades before its time. The basic argument is that greater than human AI will become a reality within this century. When that happens there will be a divide between those that want this technology and those who do not want this technology. The rest of the book explores various scenarios of war, extinction, terrorism, sabotage, and conspiracies between the "terrans" and "cosmists".

The various original terms and detailed accounts of possible actions between these two groups seem "sci-fi" at times but I think that this "sci-fi" method is used so people can better understand this enormous issue. The fact is that most people will just skip over the 'science' of the book. Personally I wish there were several more chapters describing nanotechnology, evolvable hardware, and quantum computing. These are the tools that will make "strong AI" a reality and I wish Prof. De Garis would have spent more time on these topics.

Overall, it is refreshing to see a scientist who can see the technological, philosophical, and political implications of his inventions. I would highly recommend this book to anyone interested in AI or the future.

My expectations were not met by this offering. Before finding the book on amazon I had quite a hunt to find it. Very early in the read I was first turned off by Mr. de Garis' penchant for stressing how much he simplifies things for us non-scientists. Ok, so thats kind of personal. His work is so interesting and the possibilities are astounding, but I think his expectation that a) cosmists and terrans will likely eradicate each other and b) that super intelligent machines will probably eradicate humans poorly explained. Anyway, all in all, if the reader has never foraged into this area its worth it for the exposure.

People don't agree about this book, and its hard to get a copy! Therefore, I think its probably an important book.
In any case, I think DeGaris deserves recognition for articulating the concept that AI-nanotech intelligences will one day populate the universe.
Whether these intelligences will like humans, kill humans, or simply be another version of humanity is difficult to say.
Kurzweil is more impressive on that topic.

To those of us struggling with Windows crashing our stupid computers, this book's subject, the building of super smart computers seems somewhat oxymoronic. No matter how fast you could build a machine, if it crashes several times a day it certainly wouldn't be a risk to humanity.

In another sense, the book raises a question about the possibility of Terminator like devices actually engaging in a war with humans. Can it happen? You certainly cannot prove that it cannot.

What the author is trying to do is to begin thinking about a potential problem. At that he succeeds.