Collective intelligence

Collective intelligence , also known as group or swarm intelligence , is an emergent phenomenon in which groups of individuals can make intelligent decisions through cooperation, regardless of the intelligence of the individual members .

The term has long been used in a wide variety of meanings, but it has only gained greater attention and popularity with the ability of larger groups of people to communicate through electronic media such as the Internet.

Under the term collective intelligence, very different approaches are sometimes summarized, from collective decisions by individuals that do not interact with one another or only slightly to self-organizing groups that are integrated with one another through intensive communication and thus even a higher-order individuality (a " superorganism ") can form. What they usually have in common is a decentralized, non-hierarchical decision-making structure.

Various systems theoretical , sociological and philosophical approaches exist to explain this phenomenon .

Systems theory

Francis Heylighen , cyberneticist at the Vrije Universiteit Brussel , regards the Internet and its users as a superorganism : “A society can be viewed as a multicellular organism, with individuals in the role of cells. The network of communication channels that connect the individuals plays the role of the nervous system for this superorganism. ”The swarm does not replace the network, but forms the basis. This view is in line with the view of the Internet as an information infrastructure . The meaning of the term is shifting away from artificial intelligence towards a kind of aggregation of human intelligence.

Sociological description

A certain sociological interpretation meant by collective intelligence common, consensus- based decision -making. Collective intelligence is an old phenomenon, which advances in information and communication technologies are pointing to new and intensified. The Internet makes it easier than ever before to coordinate decentralized knowledge of people and to exploit their collective intelligence.

In this sense formulated Howard Rheingold in his 2002 book "Smart Mobs: The Next Social Revolution": "The ' killer apps ' of tomorrow's mobile infocom industry will not be hardware devices or software programs but social practices." (The killer applications of tomorrow's mobile IT industry will not be hardware or software, but social actions.)

The model of swarm intelligence is assumed to have the potential to transform society and markets. Smart mobs such as the critical mass movement serve as examples .

Scientific description

Biologists have long been fascinated by how large groups of individuals, such as flocks of birds or fish, but especially the states of eusocial insect species, coordinate their behavior so that the entire group can, for example, search for food in a coordinated manner or avoid predators . Each individual has only relatively little information about their environment and only interacts with a limited number of conspecifics; nevertheless, the group as a whole makes coordinated, meaningful decisions. Through modeling, it could be made clear that the group decision is formed via feedback , in which each individual aligns his behavior with that of his neighbors and influences their behavior in turn. The behavior of individuals is similar to the structures in neural networks , for example in the brain. Social, state-building insects, despite the relatively low intelligence of individuals, are capable of even more highly organized behavior, which inspired the entomologist William Morton Wheeler to use his metaphor of the "superorganism" at the beginning of the 20th century. Since then, numerous mechanisms have been unraveled that make this achievement possible. Some coordinated interactions are made possible indirectly by changing the environment. Nest-building termites or wasp species perceive the shape of the emerging nest in their environment and react to it by adapting their own building activity. This creates the coordinated shape of the nest without the building individuals ever coordinating it directly or having an overall plan in mind. Pierre-Paul Grassé coined the term stigmergy to describe this form of coordination . The animals also communicate directly for other services. For example, ants form paths ( ant trails ) by marking the path covered with pheromones . The more ants use a path, the more attractive it becomes. This simple mechanism enables ant colonies to determine the shortest and most efficient paths between their nest and abundant food sources. Similarly, bees can not only use the waggle dance to tell their nest mates the location of food sources, but through contact with several recruiting individuals they can also find out which are the most productive sources of food. Through such relatively simple, positive and negative feedback, whole peoples can not only coordinate their behavior, but also cooperate in the fulfillment of complex tasks and thereby work together in a coordinated manner for individual tasks.

Certain aspects of the “intelligence” (better “functionality”) of an ant colony - for example processes of foraging - can be recorded in rules and simulated with computer programs. The cooperation of autonomously acting swarms of robots should also be controlled according to this model (“swarm robotics”). Similar decision-making structures are also found when human groups work together, whose coordinated “intelligence” can then not result from the intelligent behavior of the participants, but rather statistically according to similar principles. Paradoxically, if the intelligent collective behavior results only from such swarm intelligence, the more the group communicates, the worse the performance of the group.

In a certain way, a brain is also the interplay of a superorganism made up of “unintelligent” individuals, namely neurons . A neuron is almost nothing more than an integrator with a reaction threshold , more precisely, a sigmoid reaction curve . Only the complex interaction of billions of neurons, subject to specific rules, results in what we understand by intelligence.

Description in computer science

Swarm intelligence (Engl. Swarm intelligence ), the research field of artificial intelligence (AI), which on agent technology is based, is also called Distributed Artificial Intelligence (DAI). The research area tries to model complex networked software agent systems based on the model of state-forming insects such as ants , bees and termites , as well as sometimes also flocks of birds ( swarm behavior ). Gerardo Beni and Jing Wang coined the term swarm intelligence in 1989 in the context of robotics research.

VKI research assumes that the cooperation of artificial agents can simulate higher cognitive performance ; Marvin Minsky calls this The Society of Mind . An example of how these ant algorithms can be used was presented by Sunil Nakrani from Oxford University and Craig Tovey from the Georgia Institute of Technology at a conference on mathematical models of social insects in 2004 ; They modeled the calculation of the optimal load distribution on a cluster of Internet servers based on the behavior of the bees when collecting nectar.

The Knowledge Query and Manipulation Language (KQML) is often used for communication between the software agents .

In 1986, Craig Reynolds created a simulation of the swarm flight with the computer program Boids .

In addition to the research field of the VKI, swarm intelligence is also a fuzzy fashion - buzzword such as peer-to-peer (P2P) from around 2000 . While the latter began to replace the paradigm of the client-server architecture with decentralized P2P architectures, swarm intelligence is now to replace hardware-based networks .

Researchers at Princeton University , under the direction of Roger Nelson, have been studying the phenomenon of human collective perception since 1988 and have stationed measuring stations around the world for this purpose. The Global Consciousness Project collects the empirical data and compares it with the news situation in order to determine whether an event is provoking neural reactions before the news has been spread. Significant, albeit minimal, empirical evidence has been provided.

Symmetrically distributed algorithms are closely related in terms of content, but scientifically more precisely defined in the sense of computer science .

Application examples

In didactics

Learner groups are redesigned in such a way that the resources of the individual learners are used to a greater extent than is the case with traditional frontal teaching . The brain is used as a model and the learners are defined as neurons . Collective thoughts emerge on the basis of intensive interactions between the learners . This principle is used systematically in the teaching method of learning through teaching .

The Internet

Also, the cyberspace has already referred to as collective intelligence. In today's state of the Internet with its billions of largely incoherent, static documents, however, collective (in) knowledge is sometimes spoken of with a little more caution (keywords are information overload and information garbage ). However, Internet content is becoming increasingly dynamic (examples: web feeds , blogs , wikis ).

Joint review of facts

An example of the processing of a topic by a large number of Internet users is the GuttenPlag Wiki , which was installed on February 17, 2011 to check whether and to what extent the criteria for a scientific paper in the doctoral thesis of the former German Defense Minister Karl-Theodor zu Guttenberg Work were injured.

In 2009, DARPA carried out an experiment on swarm intelligence. For a prize money of 40,000 US dollars, the secret locations were to be found all over the USA, where ten red balloons were visible for a few hours on a December day. The experimental set-up should encourage a collaborative search. The experiment was successful; all ten locations were found.

Wikipedia

The rise of the Internet encyclopedia Wikipedia is widely seen as impressive evidence of swarm intelligence.

military

Under the name "Perdix" , the United States Department of Defense developed a swarm of micro-drones that are dropped from aircraft and are intended to carry out combat and reconnaissance missions with the help of collective intelligence in the target area.