Abstract

Collective intelligence is, by definition, a social view of intelligence, seen as emerging from the interaction of multiple units capable of independent behavior. Thus the cognitive nature of each collective intelligence generally derives from the properties of its underlying social interactions, and as such has been studied in as distant cases of collective intelligence as insect colonies and scientific communities. On the other hand, the most recent instances of collective intelligence, given by on-line communities like Facebook and Second Life, add complexity to the picture by letting their participants interact through the full gamut of behaviours available in real life, thus giving cognitive status to single interactions and not just to their collection. Furthermore, the digital nature of these communities, coupled with their cognitive wealth, makes possible to cross the boundary of the simple paradigm «cognition-as-interaction» by augmenting them with cognitive functions in the form of software agents dedicated to various cognitive tasks. We introduce here a few simple criteria in order to classify and compare collective intelligences according to the nature of their interactions. We then illustrate the design and the technological background for a cognitive function capable of merging digital communities with earlier instances of collective intelligence, like governments, corporations and scientific communities, thus providing a more balanced kind of collective intelligence where the «subconscious» creativity of digital communities complements the «conscious» purposefulness and goal-orientation of the earlier cases.