The Different Artificial Intelligences of Science and Wikipedia

Artificial intelligence (AI) is becoming a crucial issue for society. But the flourishing communication about AI makes it more difficult to understand what AI actually is and what role AI takes in society. From a systems theory perspective, we propose that to understand AI is to understand how societal subsystems deal with AI. To investigate the different constructions of AI we use text mining methods on the discursive structure, the semantics, and their trends over the past 20 years. We compare two subsystems – science and mass media – with getting insights from two databases – OpenAlex and Wikipedia. In science, AI is a computer science topic, consisting of four main strands: computer vision, natural language processing, robotics, and machine learning. AI as a research topic appears as a driver of scientific development that underlies machine learnization. AI as a media and broadcasting topic on Wikipedia is stable and concentrates on science, big technological firms, and science fiction. The respective pages are oriented towards the perspective of everyday life and range from personal assistance systems over sci-fi franchises to philosophical currents. We diagnose a trend from speculation to application in the attention AI receives on Wikipedia. Other systems like polity, law, or social movements do not play (at the moment) an important role in either system. Taken together, the trends point to a peculiar divergence between exaggerated expectations, everyday applications and societal debates about potentials, limits, and challenges. • Definitions, significances, and trends of AI differ in societal subsystems. • As a research topic AI strongly increase and drives scientific development. • Machine learnization of AI research: machine learning becomes over time more important in all strands of AI research. • Stable broadcast topic AI in Wikipedia contain Big Tech, (computer) science, and science-fiction. • Wikipedias attention to AI shifts from speculation to application.