International Workshop on Science, Innovation, and the Knowledge Economy

14:00～14:30 Opening and Presentation 1

Topic

Evolution of Scientific Knowledge Production

Presenter

Akira MATSUI (CCSS, Kobe Uniuversity)

14:30～15:30 Presentation 2

Topic

Preprints, Inequality, and Trust: Designing Scientific Communication for Humans and AI

Presenter

Chiaki MIURA (Graduate School of Engineering, The University of Tokyo)

Abstract

Preprints have been adopted at an unprecedented pace in the computer science community, a trend now extending beyond the physical sciences into the life sciences and social sciences. Their impact on scientific communication has become increasingly visible, particularly with the rise of Large Language Model (LLM)–based recommendation systems that rely heavily on open access content.
In this talk, I examine how scientific communication systems can be better designed for both human researchers and AI, focusing on two key dimensions: trust and inequality. Drawing on large-scale bibliometric evidence, I show that, surprisingly, while overall citations to preprints have increased since the COVID-19 surge, authors’ preference for citing preprints has declined. At the same time, citation inequality within preprints is significantly higher than in comparable journals, even after controlling for venue age and average citation levels.
These findings suggest that simple promotion is insufficient for sustaining preprints as a trusted knowledge source. Active curation mechanisms are key to redistribute attention and support more balanced and reliable scientific communication.

15:30～16:00 Coffee Break

16:00～17:30 Presentation 3

Topic

Elite Talent and Firm Productivity in the Age of AI

Presenter

Lee BRANSTETTER (Heinza College, Carnegie Mellon University)

Abstract

Artificial intelligence (AI) has advanced rapidly in recent years, and some experts see it emerging as a transformative general-purpose technology, but firms vary greatly in their ability to capture its benefits. We hypothesize that the shortage of experts trained up to the scientific frontier is an important resource constraint shaping where and how AI diffusion and innovation occur. We contribute to the study of elite human capital in this domain by constructing new data on the training of these experts and their movement from academia to industry. We use publication data to identify the most influential scientists in 10 academic subfields of AI, link these scientists to their graduate students and postdoctoral advisees who find employment in the United States, and follow the movement of these students across time and organizational boundaries.  Our data suggest that elite graduates are not disproportionately absorbed by the largest technology firms (“big tech”), but are instead distributed across a wide range of sectors. We link these experts to the firms that employ them and use Compustat data and standard panel regression techniques to explore how frontier AI talent reshapes firm productivity. Firms that hire elite graduates experience sustained total factor productivity improvements of 8--13 percent, with effects that strengthen over time. Elite graduates also enhance the labor productivity of hiring firms and their presence is strongly correlated with the generation of patents describing AI-related innovations. Event study analyses point to a causal effect of AI innovation on firm productivity. The strong association between elite graduates, AI innovation, and productivity growth points to the role these individuals may play in both generating innovations and connecting firms to sources of external knowledge.