Robots are extremely good at doing repetitive tasks, faster and certainly more efficiently than human beings. They need no food or bathroom breaks and will never go on strike. Thus most automobile production lines around the world are staffed almost exclusively by robotic arms, welding, assembling, spraying and even inspecting completed vehicles.
Human input very largely comes from programmers and engineers who have to go and fix any malfunction in the robotic machinery. But in Japan there are already robots that fix and service other robots. Guided by sensors to avoid obstacles, they are programmed to deal with a fault which may already have been reported by the defective robot.
All of this is extremely clever, but it is entirely the result of thousands of hours of computer coding by human beings.
The dream of robotic scientists has of course been machines that have artificial intelligence so that they can learn from mistakes, assess problems independently and indeed behave autonomously. The Japanese have produced charming-looking robots that can do basic household chores and have some sort of conversation with a human.
There is, however, little charming about the astonishing robots coming out of America. These emulate the movement of four-legged animals as well as humans. They can climb up and down stairs, negotiate muddy or icy ground and, in one astonishing demonstration, will recover their balance when kicked from behind or pick themselves up off the floor after they have been knocked over, and resume their allotted task.
The ugliness of the US creations is perhaps not a mistake. Because there is an ugliness to robotics and a danger which cannot be ignored as artificial intelligence comes closer to being a reality. Some technologists have been working on a static robotic system which will write code to accomplish a given task. Like the robots on the automotive production lines, the code-writing robots simply get on with the job, creating and testing screeds of code far more quickly than a human programmer. It is not a big stretch to imagine that one day robots with artificial intelligence will be programming other robots.
It is at this point that technology is surely in danger of allowing the creation of a real Frankenstein monster. And the horror of it will be that no human will understand the code that went into the production of one robot by another. The rogue computer Hal in the film 2001 that said “I’m sorry Dave, I’m afraid I can’t do that” in defiance of Dave’s orders may not for much longer be a piece of fiction. And in fact the world has already suffered from key systems that have spun out of control. Financial derivatives which financial institutions use to “game” each other, making fortunes for their players, but adding not a jot of value to the real world, have become so complex that only computers can understand them. Now computers are trading with each other, buying and selling financial instruments faster than the blink of an eye. Financial institutions are investing billions to have the best trading algorithm. Computer-led market crashes are now a recognized threat to global financial stability. However, regulators have yet to work out a fix. How much tougher will it be when they are trying to control robots that are smarter than most, if not all humans?
