Just as it took several generations before the impact of the steam engine became apparent in all sectors of the economy, computers and robots are only now beginning to reveal their full potential. It seems the answer to the question ‘Man or machine?’ will increasingly be in favour of the latter. An exploration of what lies ahead.
In 1996 the ASCI Red was the fastest supercomputer in world. It cost € 40 million, was able to perform 1.8 trillion calculations per second and was used to simulate nuclear tests. Less than ten years later another computer with the same performance capability was used not for nuclear tests but for video games. That was the Sony PlayStation 3, which cost just € 400. And the iPad now in the stores has even more processing power and greater functionality than the ASCI Red.
Moore’s law observes that throughout the history of technological progress the number of transistors in an integrated circuit has doubled every two years. This law has held true for almost half a century. What it means is that computers are more and more capable of analysing and combining vast amounts of data at lightening speeds – an activity that increasingly resembles ‘intelligence’. In recent years this has led to huge strides in areas such as machine learning (the ability of a computer to automatically refine methods and improve results with a growing amount of data), computer vision (computer analysis and interpretation of images) and simultaneous localisation and mapping (SLAM), an algorithm-based method used by self-propelled vehicles to map an unknown environment while keeping track of their location within it.
Less than a decade ago it was generally thought that the control of a vehicle could only be left to a computer under very specific conditions. Yet a few months ago a self-driving car developed by American automotive parts manufacturer Delphi travelled from San Francisco to New York with the aid of technologies such as Lidar cameras, which measure distances.
Less than a decade ago it was generally thought that the control of a vehicle could only be left to a computer under very specific conditions. Yet a few months ago a self-driving car developed by American automotive parts manufacturer Delphi travelled from San Francisco to New York with the aid of technologies such as Lidar cameras, which measure distances.
“Less than a decade ago it was generally thought that the control of a vehicle could only be left to a computer under very specific conditions”
Computers are also able to perform diagnoses without the intervention of a doctor, as technology company IBM proved last year. The company ‘trained’ a computer to monitor all relevant medical publications and analyse symptoms, test results and patient records in relation to the information. Not only was the software as accurate as the doctor, it was actually better at identifying certain features of a disease. Even natural language processing no longer appears to be beyond the scope of technology. Computers are increasingly capable of understanding human language and can repeat back what is said to them. They can even generate prose of acceptable quality. The first automatically written news item recently appeared in the US. Based on data published online by government agencies, the computer generated report was written by an algorithm created by the author.
Learning through trial and error
The exponential growth of computing power in recent years has also led to major developments in robotics: an area of technology that many feel will radically change the fabric of our society and economy in the decades to come.
Robots (a collective term for programmable machines) have existed in their most elementary form for decades. They are used in tightly controlled production environments to perform simple repetitive tasks. But we are about to witness the emergence of a new generation of robots that are capable of doing a great deal more. According to Pieter Jonker, Professor of Vision Based Robotics at Delft University of Technology, robots that help out around the house and perform basic care tasks will be a reality in the next 20 years. Jonker conducts research on robots that are capable of learning. Soon these robots will be able to carry out all kinds of tasks autonomously in complex environments.
During a discussion evening in Amsterdam Jonker recently revealed how this works. First the software that instructs a robot how to walk is removed from the robot. The robot then makes random movements and is ‘rewarded’ when it takes an efficient step. Because the robot is programmed to achieve the highest possible score, it learns to turn random movements into specific movements because these are the movements that gain the highest scores. At the same time it keeps trying out other movements in an attempt to achieve an even higher score.
Dutch professor Vanessa Evers is also working on robots that are programmed to adapt to their environment and, more importantly, the people in it. She maintains that a great deal of human behaviour can be translated into parameters and algorithms, which means that it can be learnt by a robot. So machines are moving up the DIKW ladder (data, information, knowledge, wisdom).
Unprecedented precision
Goof Hamers, CEO of Vanderlande Industries, a company that specialises in automated material handling systems, believes that developments in different areas, which are now converging, will turn (basically stupid) robots into intelligent machines at breakneck speed. Firstly, several breakthroughs in so-called sensor technology mean that it is possible to obtain 3D measurements with unprecedented precision. The resulting data is interpreted by increasingly complex algorithms and converted into actions that can be performed by robots. Hamers: “If you combine this with vastly increased computing power, you can execute highly sophisticated algorithms in real time, because that’s really what it’s all about. In our case, for example, we want robots that can decide in a fraction of a second how to pick up a particular suitcase irrespective of the shape, size or position of the suitcase.”
