Capacità di Calcolo Nazionale per la Ricerca Avanzata su AI/DL

New technologies are acknowledged as a major factor of knowledge generation. Artificial intelligence (AI), as a technology with complex nature, has the potential to impact science, innovation and consequently overall economy. This serves as an incentive for researchers to understand driving forces and barriers in producing technological knowledge that is AI. Computing capacities, data availability, and training algorithms are widely considered the drivers of modern AI research. This research aims to investigate the production patterns of AI research in science, focusing on the emergence of deep learning (DL). The overarching research questions seek to understand the concentration trends in computational capacities and the impact these trends might have on the generation of new and frontier research. We build a novel dataset on national computing capacities and AI research output and explore the impact of the former on the latter. Our findings reveal surprising trends characterizing the computational capacity landscape: although more countries are increasing their computing capacities, resulting in reduced global concentration, the distribution of these resources remains highly unequal. Interesting implications of these results arise, with this trend correlating with trends of de-democratization in AI research, where a few elite nations dominate the frontier of innovation.