This blog entry has been written by a Clickworker.
Well-known innovations concerning computers and the internet have captured the market, define everyday life and delight users. With artificial intelligence ambitious researchers want to add a new dimension to technical development. Crowdsourcing helps computer systems with self-regulated machine learning.
Watson is an intelligent, semantic computer program, which as a complicated question answering system, can process natural speech. To the question from Todd Spaletto, president of a well-known manufacturer of mountaineering gear, about what the most practical clothing for mountain hiking in rainy pre-summer weather would be, the computer answers: “There is just a three percent chance that it will rain,” and recommends that the outdoor specialist wear a “breathable fleece jacket” of a durable make.
The idea of creating an artificial creature gifted with intelligence has always fascinated people. Early on researchers dreamed of recreating human intellectual capacity in a machine, effectively reproducing the act of creation a rational being through technology. Wilhelm Schickard constructed the first calculator in 1623. From the beginnings of formal logic in the philosophy of classical antiquity, through the propositional logic of George Boole, to digital computer theory (Alan Turing), the foundation for modern research on artificial intelligence (AI) has evolved.
The notion, propagated during the Enlightenment, of man as a machine, which derives mental and spiritual processes from tangible events, has in a sense become the approach adopted by the artificial intelligence field. Based on the assumption that thinking is a processing of information, AI has set itself the lofty goal of modeling rational reasoning and the solving of problems, along with the planning of actions, sequences of movements, intellectual reactions, and speech acts with the help of computer-operated machines. Since then the development of artificially intelligent systems has made enormous progress.
Thus far robots endowed with consciousness and acting in a sensitive manner may constitute a fabulous utopia that will only be feasible in some future time. However several fantasies from science fiction (“Metropolis” from Fritz Lang, Stanislaw Lem’s “Golem” series) appear to be gradually becoming reality. Self-driving cars and drones employ artificial intelligence today, and computer systems, through controlled perception and manipulation of objects, can perform simple household activities and logistics.
Automobile corporations used robots on production lines as early as the 1960s, where there were tasks to be performed that were to some extent harmful to one’s health, like painting or welding. In the same decade the computer program ELIZA, created at the Massachusetts Institute of Technology, could effectively simulate a psychotherapeutic conversation. In 1997 an IBM computer controlled by intelligent algorithms succeeded in defeating World Chess Champion Garry Kasparov. Today’s research is mainly confined to the pragmatic questions of so-called weak artificial intelligence, which have to do with specific problems concerning the usage of intelligent computer systems. Among the variety of theoretical concepts conceived by researchers, the neuronal network models have now shown themselves to be particularly influential. Based on neuropsychology, these processes attempt to technically recreate the processing of informational stimuli in the human brain.
Machine learning, by which computer algorithms automatically increase their performance through data provided to them, finds itself at the center of the newer research on artificial intelligence. With complex search and optimization algorithms which control the computers, automatic learning processes can be generated. With crowdsourcing, algorithms can be trained, by which the crowd supplies the machines with the required data, so that just like learning through experience they can achieve ever-better solutions. Often it also makes sense to get the initial results of AI systems evaluated by the crowd regarding human logic. Artificial intelligence systems may also untilized combined with crowdsourcing for humanitarian purposes. E.g. to quickly and reliably filter out relevant images, videos and text messages from a large quantity of data. The crowd workers help the learning mechanisms by identifying photos and messages (tweets, for example), tagging them with keywords, and feeding the algorithms with important data so that they learn the essential information and can apply it to additional cases.
Computer and internet giants like Microsoft, Facebook and Google are competing to develop, through artificial intelligence, the most attractive speech-activated user interface. The virtual assistants or chat bots search the web across all sub domains, search engines, apps, and social media for the appropriate information and present it to the user in as personal a way as possible. So far no computer program has been able to mimic human intelligence well enough to fool a jury of experts (the so-called Turing Test). However because artificial intelligence research is advancing so rapidly, experts like AI specialist Elon Musk or astrophysicist Stephen Hawking recognize the danger that one day an intelligent or even super intelligent computer system could spiral out of control.
This blog entry has been written by a Clickworker.