Events

Generative AI is a transformative technology that has the potential to revolutionize the way we create and consume content. However, it raises a number of complex copyright concerns. In this talk, I will discuss the key concepts of generative AI, the current status quo, and the challenges that generative AI poses for copyright protection. I will also discuss potential technical solutions for protecting the copyright of artists in the age of generative AI.

Meeyoung Cha is an Associate Professor in the School of Computing at the Korea Advanced Institute of Science and Technology (KAIST) and an adjunct professor in the Department of Brain and Cognitive Sciences. Previously, she was a postdoctoral researcher at the Max Planck Institute for Software Systems (MPI-SWS) in Germany. Her research interest is in developing and using Data Science and AI for challenging social problems. Her research on misinformation, fraud detection, and long-tail content has received over 20,000 citations. Meeyoung Cha is the first woman in Korea to receive the Young Information Scientist Award, and she is also a recipient of two Test-of-Time Awards (AAAI ICWSM, ACM IMC). In industry, Meeyoung worked as a visiting professor at Meta(Facebook)'s Data Science Team. She is a member of the Seoul Forum for International Affairs (SFIA) and a commissioner for the Korea Copyright Commission, the Korea Customs Service, the National Tax Service, and the Open Data Mediation Committee (ODMC). Meeyoung is currently jointly affiliated as a Chief Investigator at the Institute for Basic Science (IBS) in Korea, where she directs the Data Science Group.

We provide causal evidence on the effects of language barriers on the speed and extent of knowledge diffusion by exploiting the introduction of pre-grant publications in the US. Language barriers account for almost half the diffusion lag of Japan-originating knowledge to US-based inventors, relative to Japan-based inventors. This acceleration is significant only for firms with low appropriation advantage in translation (small R&D scale, or little involvement in the Japanese market), and is larger for the diffusion of high-quality inventions, suggesting difficulties of quality-targeted translation. Thus, pregrant publication provides a significant public good for cumulative innovation through accelerated access to translated foreign patents.

The European Union’s (EU) General Data Protection Regulation (GDPR) became effective in 2018, and it puts restrictions on the processing and use of health data in clinical research. Besides limiting cross-border data exchange between EU and non-EU-based researchers, obtaining informed consent with regard to data privacy became more challenging.

The impact on clinical research is studied looking at case reports, a timely format for clinical research results. Using a regression discontinuity design taking into account that not all case reports are affected by the policy change, it is found that there was a decline in publishing clinical case reports by EU-based authors in the order of around 12 percent following the GDPR introduction. This decline is robust for various alternative model specifications and data subsets.

Follow-on analyses show heterogeneity among EU member states in terms of publication output, seemingly rooted in different institutionalized contexts, implying that clinical researchers in some countries struggle more than in others, which raises questions on equal opportunities for clinical research following a policy change that aims at standardizing data privacy in Europe. It is also found that rare and non-rare cases are affected alike.

Overall, the GDPR introduction may negatively affect knowledge diffusion, negatively impacting patient health.

Contact person: Michael E. Rose

Subscription to the invitation mailing list and more information on the seminar page.

Sharing lessons from experience (hereafter war stories) by successful entrepreneurs is a common practice in training programs for de novo entrepreneurs. Yet, war stories’ content and their effects on audiences is unknown. We examine this issue through a field-experiment in Singapore during 2019-2020, on 339 de novo entrepreneurs randomly assigned to two treatment arms: training by successful entrepreneurs sharing their war stories or training by experienced instructors imparting structured innovation strategy frameworks. We tracked venture performance until 2022. Abductive analysis of war-stories’ content revealed that successful entrepreneurs encouraged de novo entrepreneurs to focus on their ventures’ survival. Furthermore, ventures led by de novo entrepreneurs exposed to war-stories treatment experienced comparatively greater revenue growth, especially when successful and de novo entrepreneurs were similar on nationality or race.

Contact: Marina Chugunova

Subscription to the invitation mailing list and more information on the seminar page.

Robotizing to Compete? Evidence from Portuguese manufacturing Exporters
Presenter: Klaus Keller (Max Planck Institute for Innovation and Competition)
Discussant(s): N.N.

Valuation Asymmetry between Licensors and Licensees of Algorithms
Presenter: Joy Wu (ISTO) 
Discussant(s): N.N.

This paper proposes a novel framework for analyzing collaborative innovation that captures both competition and cooperation among firms, and examines the impact of private appropriation through IP rights licensing on firms' incentives to innovate and on the overall outcome. I show that when developing technology together firms compete and cooperate, and that the intensity of each force depends on their technological similarity and business model. To study the net effect of these forces in equilibrium, I focus on the standardization of mobile telecommunications technologies and use a novel dataset on the development of 3G and 4G standards to estimate my model.  I show that enforcing royalty-free clauses reduces the participation and contributions of firms, delaying the completion of the initial release of 4G by almost one year beyond the almost 3 years it took to develop.

Contact person: Rainer Widmann

Subscription to the invitation mailing list and more information on the seminar page.

Despite documented benefits of women mentoring newcomers, we argue that stable social hierarchies carry an evaluation disadvantage for senior women that transcends to impair their protégés’ trajectories. Identifying 4,556 formally mentored scientists with competitive early career funding from the U.S. National Institutes of Health (NIH), we document a citation discount of 10% on the average paper published by women- relative to men-mentored protégés. Accounting for the underlying quality and content of the work, we show that this evaluation discount carries over from papers co-authored with the mentor to protégé’s independent work. Consistent with theory of cumulative disadvantage in stable social systems, we find a widening citation differential for protégés of women versus men as careers progress. We show that the citation discount for women-mentored protégés stems primarily from their mentors’ network of citers. The mentors’ network of citers continues to contribute over 60% of citations to protégés’ work, even if the work is no longer co-authored with the mentor and published several years after formal mentorship. The stability of the gendered citation networks thus provides the ground for citation differentials afflicting protégés mentored by women relative to men. These findings raise concerns about an unbiased discourse on the best scientific contributions and about systemic limitations to women serving as mentors. More generally, the findings also hold implications for status-driven markets and the evolution of social networks.

Co-authored with Leo Schmallenbach and Karin Hoisl (both University of Mannheim)

Contact person: Elisabeth Hofmeister

Subscription to the invitation mailing list and more information on the seminar page.

This paper shows that CEO’s trust enhances innovation within firms, a novel micro-foundation for the well-known trust-growth relationship. I build a new matched CEO-firm-patent dataset covering 5,753 CEOs in 3,598 US public firms and 700,000 patents during 2000-2011. To identify CEO’s trust’s effect, I exploit variations in (i) generalized trust across CEOs’ ethnic origins, inferred from their last names using de-anonymized historical censuses, and (ii) CEOs’ bilateral trust towards inventors. Following CEO turnovers, a one standard deviation increase in CEO’s generalized trust is associated with 6% more future patents and 4-6% higher average patent quality. Changes in CEO’s bilateral trust towards inventors in different countries or from different ethnic origins have comparable effects on inventors’ patenting, controlling for CEO and other fixed effects. The effect is driven entirely by higher-quality patents, consistent with a model in which CEO’s trust incentivizes researchers to undertake high-risk explorative R&D. Furthermore, across and within firms, CEO’s generalized trust is associated with a text-analysis-based measure of firm’s trust culture computed from one million online employee reviews. 

Contact person: Marina Chugunova

Subscription to the invitation mailing list and more information on the seminar page.

Abstract follows soon.

Contact person: David Heller

Subscription to the invitation mailing list and more information on the seminar page.