Close your eyes for a moment and imagine a future populated by super-intelligent robots. Now, imagine that you are a robot
historian who is looking back at the technological lineage that gave rise to your species. You would certainly imagine a different history than a human historian would. In fact, you would probably see humans as being an important, but not overly significant, part of the evolutionary machinery that gave rise to your kind. As soon as you robots had the opportunity to remove those troublesome "meatbots" from the development loop, you probably would.
This is the chilling scenario outlined by Manuel De Landa in the introduction to his War in the Age of Intelligent Machines. Even more frightening than this piece of dystopian whimsy is the argument made by De Landa that this "removing the humans from the loop" is exactly what the military has been trying to do in the last few decades using artificial intelligence, "smart" weaponry, and sophisticated surveillance technologies.
This is, by far, the most fascinating and informative book I've read all year. In it, De Landa carefully and clearly presents a history of military technology and the gradual process of transferring cognitive responsibilities from humans to machines. There are two threads simultaneously explored here. One is this process of human-to-machine transference, the other, an introduction to the science of self-organization and chaos theory. De Landa believes that we are at a historic juncture in our relationship with machines. He sees recent work in self-organizing systems and our new understanding of the role of chaos in the dynamics of order as being significant discoveries for the future of human-machine symbiosis. We have the opportunity, afforded by personal computer technologies and the autonomous individuals who use them, to forge a relationship with machines that will help "create new forms of collective intelligence... getting humans to interact with one another in more novel ways."
War in the Age of Intelligent Machines argues that the military has been rather put out by the rapid deployment of personal computers, both in the military and in society as a whole (a society which needs to be swept up into the war machinery in times of conflict). The military has been very resistant in incorporating decentralized, "personal" tech into the military control structure because it goes against their desire for total centralized control with advisory or even executive roles assigned to AI units.
In his conclusion, De Landa makes a plea for the development of new computer communication structures that can be used to guard against the repressions of an increasingly inhuman (literally) military machine: "...in this period of time between the emergence of a new machinic paradigm and its incorporation into a tactical doctrine, new opportunities arise for the experimentalists outside the war machine. It is important to develop these opportunities in a positive way, allowing the machinic phylum's own resources to work on our side, instead of choking it with viruses and other forms of terrorist electronic activities. The same processes needed to create robotic intelligence..., and thus to get humans out of the loop, can be used to establish a computer interface that can make the dream of a people-computer partnership a reality."
(Gareth Branwyn)
ACCESS:
War in the Age of Intelligence Machines
Manuel De Landa
MIT Press/Swerve Editions
611 Broadway, Suite 608
New York, NY 10012
1991, 272 pages
graphic: Outland
My favorite headline in the newspaper during Operation Desert Storm read: "Television-guided Smart Bombs Find Targets."
- Gareth Branwyn
How can we allow the evolutionary paths of humans and machines to enter into a symbiotic relationship, instead of letting machines displace humans?
The other possibility is that expert machines will become the agents of a process of centralization of unprecedented scale and destructiveness. The pooling of expertise resources in knowledge banks could encourage the tendency to use these systems to replace human experts instead of simply advising them. In the long run expert systems could cease to be mere mechanical advisers and become endowed with executive capabilities.
