“From Nature to the Robot and Back”
Three researchers from the Science of Intelligence cluster discuss different definitions and the effects of using AI.
Apr 17, 2019
Artificial intelligence (AI) will fundamentally change our everyday lives – at least that is what scientific studies predict. The Science of Intelligence Cluster of Excellence (SCIoI) deals fundamentally with all facets of intelligence: Which fundamental laws and principles underlie the different forms of intelligence? Scientists from a wide variety of disciplines, from psychology, medicine, robotics to philosophy, cooperate in the cluster, so further honing the concept is an important milestone. “In computer science and robotics, we define artificial intelligence as a behavior in technical systems that we would call intelligent in biological systems,” says Verena Hafner, a robotics professor at Humboldt- Universität zu Berlin (HU). “Not only humans, but also individual animals, shoals of fish or robots can be intelligent. Intelligent behavior in artificial systems can be completely different from that of humans.”
Her colleague John-Dylan Haynes, a professor of psychology and neuroscience at Humboldt-Universität and Charité –Universitätsmedizin Berlin, defines artificial intelligence as follows: “In psychology, intelligence research has a long tradition. The famous intelligence quotient is regarded as ameasure ofmental performance there. However, even psychological intelligence researchers find it difficult to find a common definition. There is still no agreement on the basic mechanisms of human thought.” Sabine Ammon considers the tensions between the disciplines unavoidable: “But we can use them productively and see disagreements as triggers for new approaches,” says the professor of philosophy at Technische Universität Berlin. “In philosophy alone we have been trying to understand thinking for over 2000 years. What is discussed today as intelligence was called mind or spirit at other times.” In order to be able to cooperate at all with such different approaches to the same topic, the researchers of SCIoI have agreed in advance on a few central criteria for intelligence: it is adaptable, not only provides solutions for niche problems, but is applicable to many application situations.
“Intelligent solutions are elegant and effective; they do not work using the sledgehammer approach. Take the example of the chess computer: With a lot of computing power, all conceivable options for action are played through. However, most everyday problems are associated with high time pressure. There’s no opportunity to play through all the options,” explains John-Dylan Haynes. “It is intelligent to effectively filter out the promising solution options and only try these out.” In addition to this commonworking definition, artificial systems play a central role in the cluster. “Each new theory is recreated in a robot or a computer program. This allows us to test whether the system really behaves intelligently,” says Verena Hafner. “This isn’t a one-way street. The behavior of artificial systems helps us to ask natural systems better questions. This loop – from nature to the robot and back again – is a central working principle in the cluster.”
"Through the use of intelligent machines and systems, our cultural techniques will change"
The researchers assume that the concept of intelligence is not static. “Human intelligence always arises from the exchange of culture and technology. Through the use of intelligent machines and systems, our cultural techniques will change, and so will what we call intelligent behavior,” says Sabine Ammon. Researching the effects of the growing share of artificial intelligence in our everyday lives is also one of the goals of SCIoI. “According to a research paper from 2013, 47 percent of all jobs will potentially be replaced by computers. This study was quoted in all media at the time,” recalls John Dylan Haynes. But a closer look revealed that the study was too undifferentiated and the assessment had to be strongly put into context. Today, much lower quotas are assumed, and some studies even see a net increase in jobs.Over the course of technical developments, some professions have always become obsolete. Nowadays nobody mourns that there are no more suppliers for ice blocks, no more gas lamp lighters, elevator boys, or telephone operators,” emphasizes Haynes. “Professions that require manual dexterity or social interaction and empathy, such as hairdressers and teachers, will be difficult to replace with robots in the foreseeable future. What is decisive, however, is that we succeed in achieving a harmonious partnership between humans and machines. The interaction between human and artificial intelligence has great potential.”
Researchers see the current changes as an invitation to take action. None of us are capable of anticipating social change. We can only reflect on future developments from the experience of the present. The extent to which these predictions become reality is also in the hands of society. “The topic of artificial intelligence polarizes people: utopian promises collide with apocalyptic fears. In addition, there is the feeling of a rapid acceleration of technological changes and the concern of being confronted with fait accompli,” says Sabine Ammon. “I am optimistic that our society is robust enough to find appropriate responses to the new challenges. One key is certainly the time factor. “It will require social dialogue and sufficient time for learning processes and the design of new technologies in line with our values.”
This text originally appeared on February 22, 2019, in a Tagesspiegel newspaper supplement published by Freie Universität Berlin, Humboldt-Universität zu Berlin, Technische Unversität Berlin, and Charité - Universitätsmedizin Berlin.
Science of Intelligence
At the heart of the joint Cluster of Excellence of Technische Universität Berlin and Humboldt-Universität Berlin is a better understanding of intelligence in all its facets: Which fundamental laws and principles underlie different forms of intelligence – be it artificial, individual, or collective intelligence? The 25 researchers from various disciplines want to use their findings to create new intelligent technologies.