Like the Ronald Reagan Star Wars program, the SKI idea was too futuristic for its time. But today, when we see striking progress in the creation by the military of artificial intelligence and independent robots, it makes sense to return to this half-forgotten program and ask ourselves: are we ready to live in a world of interconnected electronic brains of killer machines? And one more, perhaps useless question. If we want to stop this, are we too late with our desire?
“The possibilities are truly staggering ...”
If the technology of the new generation develops as we expect, unique new opportunities will appear for the use of computers in military affairs. For example, instead of adopting simple guided missiles or remote-controlled aircraft, it will be possible to launch fully autonomous land, sea and airborne vehicles capable of performing complex and very diverse reconnaissance and strike missions. The possibilities are truly staggering, and they say that the computing technology of the new generation will radically change the nature of future conflicts.
This is an excerpt from the obscure document presented in October 1983 to the congress. It sets out the objectives of the new Strategic Computer Initiative. And like everything else, what DARPA did before and after that, this program turned out to be extremely ambitious.
The concept of the Strategic Computer Initiative was embodied in a completely new system, the development of which was headed by Robert Kahn (Robert Kahn), who led the department of information processing techniques at DARPA. As reported in the Strategic Computing book, published in 2002, Kahn was not the first to get an idea of this system, however, “he was the first to outline the concept and structure of the future Strategic Computer Initiative. He launched this project and defined its content at an early stage. SKI found its own life, it was led by other people, but it retained the influence of Kahn ”.
This system was supposed to create such a world where independent vehicles not only collect intelligence information about the enemy around the world, but also have the capabilities with deadly precision to strike from land, sea and air. SKI was to become a global network connecting all aspects of the US military-technical potential - a potential based on new and incredibly fast computers.
But this network was intended not just for cold and impartial automated information processing. No, the new system had to see, hear, act and react. And most importantly, she had to understand, and without any prompting from the person.
Economic arms race
The origin of the SKI is often associated with technological competition that has arisen between the United States and Japan in the early 1980-s. The Japanese wanted to create a new generation of supercomputers, which were supposed to form the basis of an artificial intelligence system. Combining the economic power of the Japanese state and the new possibilities of the microelectronics and computer industry in the country, they began to create a fifth-generation computer system to achieve their goal.
The goal was to develop incredibly fast computers that allow Japan to break away from other countries (primarily from the United States and the Silicon Valley that was born there) in the race for technological superiority. To accomplish this task, the Japanese gave themselves 10 years. But no matter how they accelerated their machines, they, like the Americans, could not make computers "smarter" at the expense of powerful artificial intelligence.
Japanese aspirations frightened many Americans. They were worried that America was losing its leading technological position. These fears were largely ignited by Edward A. Feigenbaum and Pamela McCorduck’s “The Fifth Generation: Artificial Intelligence and Japan’s Computer Challenge to the World” published in 1983 (“Fifth Generation. Artificial Intelligence and Japanese Computer”). challenge to the world), which has become a must-read literature on Capitol Hill.
In order to popularize the ideas of SKI among the American people and business circles, DARPA insisted that the goal of the initiative from the very beginning was only to advance the country's economic interests. The side effects of this technology were to create new incentives for the US economy, as reported in the DARPA planning document:
The computer technology of the new generation will come to the consumer electronics industry, creating an internal market for the application of machine intelligence.
Appeal to the private sector and the university system also had to provide assistance to the most intelligent and talented in carrying out the tasks of the Office of Advanced Research and Development:
Equally important is the transfer of these technologies to the industry to create a cadre base of engineers and linkers of systems familiar with computer science and machine intelligence technologies, who today study and work in leading university laboratories, as well as the application of these new technologies in the range of products of various companies. To this end, we will fully use the regulations and rules of public procurement relating to the protection of corporate production information, trade secrets, patent rights, licensing standards and license fees.
And what is the conclusion? The government gave guarantees to the private sector that the technologies developed will not be transferred to competing companies.
But economic competition with the Japanese, albeit an important driving motive, caused only secondary concern among politicians entangled in the upheavals of the Cold War. The Republican Party hawks were most concerned with military construction and military buildup. Many of them believed that the military threat coming from the Soviet Union was most important. And the Strategic Computer Initiative had to eliminate this threat.
Connection with Star Wars
The launch of the SRS program and DARPA terms of reference, which appeared in 1983 and 1984, sparked heated debates in the scientific community - the very one that ultimately benefited from funding under this project. Someone expressed doubts about the possibility of implementing ambitious plans to create advanced artificial intelligence. Someone was worried about the fact that with the creation of artificial intelligence for military purposes, the terrible era of independent armies of robots would begin.
And it was a well-founded concern. If the goal of Star Wars (the popular name of the Ronald Reagan Strategic Defense Initiative and the popular political football of the time) is an automatic or semi-automatic response to any nuclear missile threat from the Soviets, then it would be ridiculous not to include it in a larger system of truly sensible machines. The tasks of the two projects, not to mention the institutions that developed them, overlapped and overlapped too much to be a coincidence, although each insisted that it was a coincidence.
From the work of Chris Hables Gray, written in 1988:
The Star Wars Combat Control System, which is perhaps the most complex and large-scale software project in its entirety. history, conceptually (but not administratively) is part of the Strategic Computer Initiative. Making a scientific breakthrough in computing, a much-needed SDI, is the key goal of the SKI.
If you ask someone who has worked in the leadership of the SKI program, then you will be persistent in saying that the Strategic Computer Initiative had nothing to do with Reagan’s dream of Star Wars. But people from the very beginning of the implementation of the SKI conducted a connection between it and the IDF. Partly such associations arose because of the similarity in the names and because they were given the names by one person - Robert Cooper (Robert Cooper), who served as director of the Office of Advanced Research and Development of the US Department of Defense from 1981 to 1985. Or maybe people saw the connection due to the fact that the computer interface systems developed for the SRI were quite logical to be used as an application for the space missile defense strategy.
The use of strategic computer technology on land, at sea and in the air
The overall ICS scheme prepared in 1983 for the year outlined the objective of this initiative. The goal was clear and understandable: to develop an extensive base of artificial intelligence technologies to strengthen national security and economic power. But in order to reach it, the congress and the military departments that were supposed to use the SKI and its advantages in the future, had to see this system in action.
SKI had three hardware implementations that were supposed to prove its combat potential, although by the end of the 1980s it was planned to develop even more such systems. In the forefront of the technical development of the SKI were autonomous ground vehicle ALV, "assistant pilot" and the aircraft carrier combat control system.
These funds were planned to be equipped with incredibly advanced computers, which were designed at Cambridge BBN, best known for their work in creating the first version of the Internet. Computers made it possible to achieve breakthrough successes in areas such as vision systems, language comprehension and navigation. And these are the most important tools for creating an integrated man-machine military force.
A car without a driver - 1985 year
The most sinister appearance of the product, which emerged from the depths of the SKI, was an autonomous ground vehicle ALV. This eight-wheeled car without a driver was three meters in height and four in length. It was equipped with a camera and sensors that were mounted on the roof and controlled the movement of the car, being its “eyes”.
The company Martin Marietta, united in the 1995 year with Lockheed Corporation, which resulted in Lockheed Martin, won the tender for the creation of an experimental autonomous ground vehicle in the summer of 1984. For the three and a half years of the implementation of the SKI program, she should have received 10,6 million dollars (adjusted for inflation, it is 24 million) plus 6 million in addition if the project meets certain benchmarks.
The October Science 1985 issue of Popular Science has an article about the trials that were carried out at Martin Marietta’s secret test site southwest of Denver.
The article's author, Jim Schefter, described the test site at the test site as follows:
The box-shaped blue and white car moves slowly and steadily along a narrow road in the Colorado Valley, not venturing to go far from the centerline. The only window that looks like a Cyclops eye is installed on the front side of the car, but the driver is not visible there. It moves cautiously, almost slunk, which seems somewhat out of place for this eight-wheel, three-meter vehicle height. Although it has three snarling diesel engines installed, the car travels slowly, at a speed of less than five kilometers per hour.
After about a kilometer, the clumsy car stops. But nobody leaves it. Just in the car there is nobody - only one computer. Using a laser and a video camera as an eye, an experimental, but already very complex artificial intelligence program leads the car on the road without human intervention.
DARPA joined forces with Martin Marietta and the University of Maryland, which did a lot of work to create a vision system. Such an association seemed important to ensure success in developing a land vehicle.
Creating a video system for an autonomous car turned out to be incredibly difficult. It can be misleading light and shadow, and therefore the degree of reliability she was not enough. In the afternoon she discovered the roadside without problems, but because of the evening shadows at sunset she could easily slip into a ditch.
Any changes in the environment (for example, dirt from under the wheels of another car) also confused the vision system. This was unacceptable even under test conditions at the site. If a machine does not cope with such simple obstacles, then how will it act in difficult and unpredictable combat conditions with countless variables?
By November 1987, the autonomous ground vehicle had been significantly improved, but by the end of the year it was virtually abandoned. Although the car was rather primitive, some DARPA experts thought it was too quickly discarded.
As a result, she could not overcome her unpreparedness for battle. As Alex Roland points out in his book Strategic Computing, “one officer who did not understand the purpose of the ALV program at all complained that the car was completely useless in military terms: very slow and white, which makes it an easy target on the battlefield. " In April, 1988, the Advanced Research and Development Authority, officially discontinued work on it.
R2-D2, but in real life
The second practical embodiment of the Strategic Computer Initiative was the “assistant pilot”. The developers imagined it as an invisible robot R2-D2 - a smart companion who understands the simple language of the pilot. This assistant could, for example, detect the enemy’s target and ask the pilot if it was necessary to destroy it. Something like "The best shooter" in the company of Siri's personal assistant from iphone.
In this scenario, the final decision was up to the pilot. But his assistant had to be smart enough not only to know who was asking questions, what he was asking, and how to ask questions himself. He had to understand why.
Here are lines from the SKI planning document:
A huge amount of information falls on a pilot in battle, he is constantly in touch, and on this basis must make decisions on which his life often depends. He also has a huge number of buttons, switches and buttons on the panels and control knobs that require great clarity and accuracy. Each of the hundreds of parts is designed for its own, well-defined and important goals, but the underlying technologies are far ahead of our ability to competently and intelligently establish the interaction between these components and the pilot.
And it was here that the Advanced Research and Development Administration decided that it needed its own Skynet. New features of the fighting, associated with the rapid development of military technology, demanded a clear interaction between the machine and man - and this was the key to success in battle. The pilot still pressed the buttons, but these computers had to think for him at least half. If humanity does not have time, it is necessary to connect to the operation of the machine.
The program "assistant pilot" was not covered in the American press in the same amount as the autonomous ground vehicle. Perhaps this was due to the fact that it was much harder to imagine than a huge tank driving along the road without a driver. But if you look at today's speech recognition technologies, it will become clear what all these studies led to by the “assistant pilot”.
Invisible Robot Advisor
The command and control system became the third practical embodiment of the SKI program, designed to prove its expediency.
Here is what Roland writes about this in his book Strategic Computing:
In the naval combat control complex SKI, an artificial intelligence system had to draw conclusions about the enemy and its troops, combat strength and order of battle, taking into account uncertainties, work out strike options, carry out simulation modeling in order to evaluate these options, develop operational plans and present arguments and explanations.
The combat control system was essentially the brain of the entire operation, and for this reason it was kept secret, unlike the ALV. Driving on the road without a driver, the robot can frighten many. Invisible robot with an invisible finger on the nuclear button? Well, to publish press releases on this topic is unlikely to anyone wants.
The combat control system was designed as a software application specifically for the Navy. (An autonomous ground vehicle was created specifically for the ground forces, and an “assistant pilot” for the Air Force.) But in fact, it was just a cover for a more universal system. All these technologies were planned to be used in perspective where they will be most needed. The speech recognition program developed for the “assistant pilot” was planned to be used in all types of armed forces, and not only in the Air Force. And the combat control system should have been suitable for everyone - except, of course, the enemy.
Build Skynet together
All the various components of the Strategic Computer Initiative were part of a larger hypothetical system that could radically change the nature of war in the 21st century.
Imagine a global wireless network that controls many other subordinate networks in the US military. Imagine how armies are robotic tanks they talk with swarms of drones in the sky and submarines without crews at sea - and the interaction between them is much faster than any human commander could do. Now imagine that all this is much more complicated with nuclear missiles awaiting launch into space.
The concept of Strategic Computer Initiative was incredibly brave, and at the same time, a bit unusual, if you think about how far it could lead us. The logic of the further development of artificial intelligence and the worldwide network of killer machines is not difficult to imagine, if only because we have seen it in books and movies without a score.
Future of war and peace
The strategic computer initiative at the beginning of 90-s was finally destroyed by the realization that it was simply impossible to create powerful artificial intelligence like the one that DARPA imagined. But if all these technologies and technical innovations developed in 1980's seem strangely familiar to us, it is because the media speak and write about them at the beginning of the 21st century.
Vision systems from an autonomous ground vehicle are embodied in Atlas-type robots from Boston Dynamics. We see that the speech recognition system type Siri from the "assistant pilot" is used in the US Air Force. A stand-alone machine is experiencing Google, along with many other firms. All these are technologies of future wars. And if you believe Google, it is also the technology of the world of the future.
Google Inc. recently bought Boston Dynamics, and it caused a great surprise among those who are concerned about the future with armies of independent robots. Google claims that Boston Dynamics will fulfill all of its old contracts with military customers, but will not enter into new ones.
But whether or not Google will accept orders from the military (which is quite possible, since they can do it secretly, using funds from their “black” budget), there is no doubt that the line between civilian and military technology has always been blurred. If Boston Dynamics will never again work with organizations like DARPA, but Google will benefit from research funded by the military, then you can probably say that the system works.
The military achieved what they needed by promoting research in the field of robotics through a private company. And now the results of these military technologies will make themselves felt in our everyday civilian life - as well as many other technologies, including the Internet.
In truth, this article only contains a drop in the ocean from among those plans that the Advanced Research and Development Authority bore under the SKI. Hopefully, by continuing to explore yesterday's promising concepts, we can gain some historical experience and better understand that our new achievements did not come from the air. They cannot even be called innovations. This is the result of years of research and billions of dollars in allocations that have been mastered by hundreds of organizations, both public and private.
Ultimately, the Strategic Computer Initiative was eliminated not because of the fear that it could bring to our world. It is just that the technologies for its implementation did not develop quickly enough - this also applies to artificial intelligence and autonomous vehicles. But for twenty years since the collapse of the SKI, all these developments of smart machines continued.
The future with very intelligent and interconnected robots has almost become real. We do not have to love him, but we cannot say that no one warned us about him.