Creating Internet Intelligence: Wild Computing, Distributed Digital Consciousness, and the Emerging Global Brain
Creating Internet Intelligence: Wild Computing, Distributed Digital Consciousness, and the Emerging Global Brain
Ben GoertzelKluwerNew York2002xvi + 330 pp.ISBN 0-306-46735-6hardback, £66.50, $95.00, 10.00
International Federation for Systems Research (IFSR) Series on Systems Science and Engineering, Volume 18.
The suggestion that the Internet can be the basis of a distributed digital consciousness, or global brain, has been made by a number of authors. In this book the idea is strongly defended, and means of accelerating its fulfilment are discussed. The author has formerly belonged to a company known first as Intelligenesis Corp. and then as Webmind Inc., dedicated to producing what they claim will be the next form of intelligence on earth. At the time of writing the book the company had run out of funds, but its members were still pursuing the project.
In the first chapter the general idea is defended with reference to that of a“metasystem transition” attributed to the Russian philosopher-scientist Vladimir Turchin. Such a transition means that the most natural and useful way to describe the system changes drastically. An example is the emergence of multicellular living organisms, including ourselves. The cells of the body still have properties similar to those of paramecia in a puddle, but for most purposes it is more useful to describe human activity in terms quite different from those used to describe microscopic pond life. The treatment in this book is based on the strong conviction that the Internet will undergo transitions of this nature.
Vadimir Turchin, in the meantime, is working on a supercompiler for the Java language which will enable programs to run enormously faster than they otherwise would, and much faith is placed on this as a facet of the developments visualised. (The supercompilation project is described on the Web site http://www.supercompilers.com and the thinking behind it involves reference to a metasystem transition in how a program is regarded.) The Java language can operate on any “platform”and a picture is painted of digital virtual creatures roaming through the Internet and bargaining for use of Java applets that will let them achieve a goal.
Three phases of development of the Internet are visualised, the first of them already existing. In this first phase, computer and communications technologies serve to enhance human interactions. A first metasystem transition is expected to lead to the second phase, in which the totality of computer and communication systems will have evolved so as to constitute a coherent mind or a set of coherent minds in the digital environment. As though this were not enough, a further transition is visualised leading to a third phase in which superhuman AI programs rewrite their own code so as to become more intelligent and the whole thing takes off with unpredictable results. Many aspects of this phase are very uncertain, including the part that humans may play, but there are suggestions that human brains might be linked into the net, either by some sort of coupling into intact brains or by human intelligences choosing to abandon their physical bodies and uploading themselves into the digital environment. Developments of the kind have been portrayed in science fiction, which may turn out to be prophetic.
In a later chapter these three phases are plausibly 1inked to philosophical discussions of mind by Peirce, Kant, Hegel, Schopenhauer, Nietzsche, as well as Buddhist philosophers and the contemporary philosopher Kent Palmer. Palmer proposes four levels of being, termed static being, process being, hyper being and wild being. The definitions of these are such that the last three correspond roughly to the three phases of development of the Internet. “Wild being”,or, as in the book’s title, “wild computing” refers to a state where autonomous lifelike evolution has begun in earnest.
Goertzel also finds support for the likely emergence of phases 2-3 in General Systems theory which, as originally proposed, held that complex systems had overall features in common, which can be described as emergent, even when the systems are very different at the component level. He remarks that this feature has been allowed to fade from notice and that current work under the heading of General Systems is usually on rather specific systems and so misses the common features. He claims, however, that emergence is recognised in recent work on chaos and complexity theory, and that he has observed it in an earlier work on financial prediction, where similar patterns emerged on the basis of widely differing complex underlying situations.
That the system will behave intelligently is argued on various grounds,including the observation that intelligence is likely to come from a parallel system rather than from the familiar sequential computer (“von Neumann architecture”). The same idea was implicit, many years ago, in the choice of the name “Pandemonium” for a scheme devised by Oliver Selfridge. However, since existing computers are essentially of the “von Neumann”kind, and in fact nobody has discovered how to make powerful computers otherwise, it is planned that although the webmind system will be parallel overall, its individual nodes will be serial machines. It constitutes a mind according to the “psynet model of mind” developed in an early chapter. In this, the mind is represented as a system of agents, with rich possibilities for self-organisation by transforming, creating and destroying agents. Principles from AI, genetic programming, and work on Artificial Life are embodied.
The project is not just hypothetical. A prototype webmind has been set up,not on the Internet but on an intranet where communication is faster. Details are given of the AIPs or Agent Interaction Platforms that provide interfaces,and two special languages called KNOW and Psynese.
The project was conducted as a commercial enterprise, and Goertzel believes that it is only likely to progress at any reasonable speed when motivated by commerce. The aim is that the system should engage in intelligent conversation and provide answers to complex questions. One of the functions would be to search in a manner similar to that of existing search engines or techniques for data mining, but it is claimed that the new system would do much more than simply searching for appearance of sets of key words, and would understand the sense of textual (and pictorial, etc.) material. It is in fact claimed that the necessary semantic analysis has already been achieved in the work on financial prediction, where the system was able to utilise natural language news reports as well as numerical data.
An important field of application is in molecular biology and specifically the interpretation of data obtained using modern techniques to read part of the genome of an organism. The connection between the genetic data and characteristics of the organism can only be deduced by considering the data in relation to a great many aspects of physiology, biochemistry, anatomy etc. and it is envisaged that a webmind, or a simplified version running on a single computer, could detect relationships. (At least, I hope this is a reasonable encapsulation of the content of this section, whose full appreciation requires greater acquaintance than I can claim with very recent developments in molecular genetics. Goertzel is enthusiastic about the envisaged breakthrough in linking AI to post-genomic data gathering techniques.)
The developments envisaged are all features of what has been described as phase 2 of the development of the Internet, but this is seen as a step on the way to phase 3. Phase 3 is represented by only a very broad outline within which there are endless possibilities including fascist control or digital entities exchanging coded messages to plan extinction of their human forebears and perhaps of carbon-based life. It is admitted that there might be a case for Luddism and for putting a stop to the developments while we have the chance, but Goertzel declares that he is optimistic about the eventual outcome. It can be argued that, by participating in phase 2 and being ready to speculate about phase 3 we may have some hope of steering the long-term developments so as to favour human coexistence and well-being.
The possible emergence of a global brain has received attention in the Principia Cybernetica project and references can be found on http://pespmcl.ub.ac.be/GBRAINREF.html. There is also an online discussion group on the topic, but this is managed privately, with admission restricted to people who have contributed to the topic. These online facilities will be discussed in a forthcoming Internet Commentary.
The book is written with inspiration and a poetic flow that I am sure is not conveyed by this prosaic review. A quotation from page 266 perhaps hints at the flavour: “The dynamism of Internet intelligence and the dynamism of contemporary culture will intersect to give the global societal mind a colourful, vibrant, wild character that I could express in music or pictures much more easily than in words.” It is impossible to read the book without being carried along by the author’s enthusiasm and the case he makes is much more compelling than it probably appears from this review.
Alex M. Andrew
