Thanks to the ever increasing miniaturization... 

...in the fields of electronics and micro mechanics there already exist robots practicing brain and bone surgery with absolute precision (like the Robodoc at the Pittsburgh Shadyside hospital, always worth a call!). Long distance-controls enable robotic explorers like Dante to creep into a volcano in Alaska, while the scientists sit around in California enjoying the pictures. But this text isn’t supposed to be a glorification of technology, there’s plenty of that crap around. While robots are able to cope with specific problems, they cannot really handle a dynamic world and the step towards enhanced machine autonomy requires logical calibers of a different kind. Since the 60s the AI-researchers are stuck in an eternal spasm of intelligence. The first efforts to create artificial intelligence date back to the early 60s just when Armstrong was plodding around in the Mare Tranquilitatis. Modernity had invested its victory prey into moon rocks and everything seemed possible. The first artificial brain was foreseen to exist at the turn of the century; in fact I do not even trust my toaster.

The main problem...

...is the almost infinite complexity of the human central nervous system. A robot can recognize a misdimensioned machine-built construction unit down to the last little micrometer. Precision in a fully controlled environment is its (life’s) work. The human processing system, however, can seize almost any environment in no time. The supercomputers cannot keep their pace, and the neuron scientists don’t even know for sure what their mistake is.

Chuck Thorpe, Carnegie Mellon University, thinks that the cycle “perception - processing - action” does not yet function automatically, because it is mainly “perception” which does not work. The human brain is excellent in setting up mental models of the external world and to modify them in constant reciprocity with the environment. At the moment the most successful labs are able to equip their test programs with the mental capacity of a one-year old. The robot can teach itself to balance, it can walk upright, and it can recognize the difference between a dark shadow and a hole in the ground. But these results are just not enough to rest on one’s laurels. There’s not even enough time for micro sleep. There’s lots of work to do for computer scientists and neurologists. “Adaptive”, non-linear neuronal nets may be a ray of hope, but texts and meetings concerned with this topic tend to get lost in a discursive network of prognoses of trends, speculations and analysis of the de facto technical situation. Sometime something will occur that will change us somehow. The question is when and who believes in that. This will mainly be large companies.