Bridging robotics/AI and real-world labs: A quantitative approach based on mining German newspaper articles

Both, robotics/AI (RAI) and real-world labs (RWLs), are current topics in public innovation promotion policies, but are mostly treated in isolation. While RAI has a focus on a specific technology to serve society, RWLs address the institutional context including experimental learning of governments and societal perspectives. We are particularly interested in the interface between RAI and RWLs and the way media are reporting on these two domains. This reflects key aspects of the social debate in relation to RAI and RWLs. We base our analysis on the understanding that technology development and diffusion ultimately depend on institutional arrangements that are developed alongside or in lieu of market arrangements and also reflect societal needs. This paper uses quantitative text analysis to examine 3,800 German broadsheet newspaper articles in the period 2016–2023. We use Structural Topic Modeling (STM) with publication date and sub-corpus source as covariates to trace topic dynamics and topical prevalence contrast. We show that the dominant topic has changed over time from RAI (“Machine Learning and AI Development Methods”) to RWL (“Real-World Labs for the Energy Transition”). We identify bridge topics and argue that these are diverse and include philosophical and legal considerations, public funding and specific application areas for robots, e.g. in schools. As indicators to identify the interface between the two domains (RAI, RWL), we propose a combination of topical prevalence contrast and eigenvector centrality and the use of psycholinguistic attributes to evaluate the topics. These elements could be broadly used to exploit possible complementarities for government experimental learning and when designing “smart regulation” which targets several fields simultaneously.