Sunday, August 16, 2015

Minefields: Too-smart robots

Asymptotic Technology is nice if you can get there. It is the gateway to spaceflight and to establishing the social conditions necessary for a very long-lived civilization, one easily able to handle long-lasting projects like space travel. But as noted elsewhere, there are both roadblocks and minefields on the way to Asymptotic Technology. Roadblocks are obstacles that can slow down or even stop progress toward it, and minefields are situations it might lead to that will damage the civilization, perhaps so seriously that it stops its progress and regresses. A restart might not be possible in some situations, such as exhaustion of resources.

One minefield that is beloved of science fiction writers is the too-smart robot, one which is not just too smart, but which has a philosophical bent and decides, like Descartes, “I think, therefore I am”. The idea is that the robot goes out of control of its makers, and somehow obtains enough power to damage their civilization, or even eliminate them. Is this a realistic concern for an alien civilization en route to asymptotic technology?

Science fiction writers like to be dramatic, so the transition from good robot to bad robot is sudden. The suddenness leaves the makers of the robot unable to respond quickly enough to react. Somehow the robot amasses enough control to threaten the society or some part of it, and the story unfolds to a pessimistic or optimistic conclusion. This is wonderful for a novel, but in reality changes to programs are gradual. One of the key points of this blog is that the aliens from the alien civilizations we are discussing are intelligent, and it goes without saying, observant. If a robot or controlling computer started showing signs of acting out of design, the robot or computer would be never sold or put into production or whatever. It is unlikely that there would be a sudden tremendous change in capability. Rather, there would be incremental improvements, and if any of them showed signs of slipping out of control, or operating other than according to design, fixes would be put in.

This means that artificial intelligence will be at the service of the alien civilization. In the minefield post, it was considered that individuals who were not in tune with social norms might somehow take actions that were disruptive. One action that could be contemplated would be for the sociopathic individuals to try to take advantage of the power multiplier of robots, and use them to aid in disruption. One example of this and there would be a reaction. Some protection would be built in. If that was defeated, and another individual took advantage of a loophole in the protection to do something disruptive, there would be a response. Furthermore, at this point in society, remedies for sociopathy and preventive steps would also be falling out of the research done to achieve asymptotic psychology. So both of these currents are flowing toward the same end-point: robotics without the potential for a science fiction drama.

The same process would be used for these problems as would be used for other programmatic problems with robotics and computer controllers. If a new model of a robot was being tested in manufacturing, and it destroyed some percentage of the items being manufactured, the robot would not be released, but changes would be made and testing resumed. Anyone who has worked in engineering facilities here on Earth knows that some engineering firms take shortcuts, and do premature releases, and create end-user problems. The same group of people also know that such choices are short-sighted, and the firms that do this do not usually prosper. In situations where monopolies exist, the results might be different as the end-users have no alternatives, but if there were tragic accidents arising, some reaction would be made. Minor annoyances, frustrations, and productivity losses are one thing, but the disruption caused by out-of-control robots is vastly different. Alien societies might be remarkably differently organized, but the basic principle of a necessary response to a serious failure comes from the very elementary and indubitable assumptions of an intelligent set of citizens who can observe and react. It doesn’t matter whether they have eight arms or two tentacles, they still will have the intelligence to observe the malign operation of a badly programmed, either inadvertently or deliberately, robot.

A minefield is a set of problems which might arise if no care is taken to avoid triggering the mines. One mine might certainly be a robot or a controlling system of some major piece of infrastructure, like the transportation system of a city, which starts to act in a destructive or disruptive way. How could it be that no response was made to these acts? Did the population lose all ability to respond to problems, and they depend on other robots to fix everything. This sounds like the citizens descend so deeply into the Happy Life Great Filter that they abandon control of their systems, the ones which maintain the operations of the city in which they live. This would be the root cause of the problem, not the acts of some malcreants who obtain access to the programming of a vital computer control system. So instead of there being a minefield object relating to out-of-control robotics, there is a possibly valid minefield object of loss of ability by the population to be in charge of their own planet and their city and its systems. This might happen if the population lost its intelligent members, but how could that happen? Intelligence is part of the prescription for the gestation of new citizens. It could be they lost interest in performing any action even for self-preservation or maintenance of the city, but how could that happen? Education is a subject that will have been fully researched and applied to the newest generation of citizens. Where is the gap that cannot be filled? Where is the hidden flaw that several thousand years of working out problems will not find and will rise up to destroy the civilization?

The answer is that time cures all ills. While this is a bromide, it is relevant in a new way for alien civilizations. Time, meaning generation after generation of citizens living in remarkably similar surroundings, will reveal all the potential problems as well as the fixes for them. Out-of-control robots and controllers do not seem to be a problem that will prevent an alien civilization from continuing to live and to develop star travel if they choose to.

No comments:

Post a Comment