When will the machines take over ?
The achievements of information technology are impressive. ICT has become an indispensable part of our society. However, the predictions of the end of the age made by some experts in the field of artificial intelligence (AI) are problematic. A plea for logical thinking and common sense.
Steve Wozniak, co-founder and technical mind behind the creation of Apple, feeds his dog fillets in the hope of creating goodwill with the robots of tomorrow. Joachim Schoss, founder of Scout24, has bought a haven in the southern hemisphere. And Elon Musk considers artificial intelligence (AI) to be "perhaps the greatest threat to humanity".
Experts predict that humans will hand over the scepter to intelligent machines and robots. The late physicist Stephen Hawking speculated that a highly developed AI could develop and reproduce itself and ultimately herald the end of humanity. And Jür- gen Schmidhuber, renowned AI expert, predicts: "AIs will populate and reshape the solar system by means of self-replicating robot factories." Technological singularity is what experts call this point of no return. The time horizon mentioned is just a few decades.
These prophecies of the end times are all based on an immense overestimation of the potential of machine intelligence. Since they trigger fear and anxiety in many people, these predictions must be resolutely opposed.
Fascinating ICT achievements
Information and communication technology is shaping today's world to a degree that was hardly imaginable just a few decades ago. In some disciplines, computers are now leaving humans behind. Back in 1996, IBM's chess computer Deep Blue won the first game against Garri Kasparov. In 2011, the AI software Watson followed with its stunning victory over humans on the quiz show Jeopardy. And in 2016, Google's AlphaGo software beat the world's best players at the Asian game of Go.
But even in domains that are more relevant to the well-being of people, the progress made in the past decade has been as immense as it has been fascinating. Prominent examples include medical expert systems and autonomous driving. Great advances have also been made in text and speech recognition, image analysis and language translation, which have had a lasting impact on our private and business lives. How effortlessly and conveniently Siri, Alexa and Co. show us the way to the world's knowledge never ceases to amaze.
These applications use neuronal networks that are based on biological models. They solve a given task via trial and error. They optimize their behavior by evaluating successes and failures. In doing so, the machines combine the high-performance evaluation of large amounts of data with machine learning. The computer operates at a higher level of abstraction by searching for characteristics and patterns and developing suitable rules based on vast amounts of sample material (deep learning). The machine no longer needs to be hard coded.
And where is the intelligence?
However fascinating the capabilities of artificial intelligence may be, they have very little in common with intelligence in the human sense. IBM's Watson doesn't even begin to understand the meaning of the questions in the Jeopardy quiz. We observe the same when we ask Google a somewhat more sophisticated question. But even with translation services, AI does not operate on a human, logical-intelligent level, but on a statistical level.
