Current Research Projects
Selective revealing in open-source artificial intelligence
In the fierce competition for dominance in the field of artificial intelligence (AI), exclusivity in data and derived models has been identified as a primary source of competitive advantage by researchers and practitioners. Following traditional private investment theorizing, such exclusive resources should be preserved and capitalized upon. Most prominently, OpenAI exemplifies this approach, having predominantly focused on developing proprietary closed-source models. Seemingly contradictory, not all companies adhere to the aforementioned rationale, with some opting to reveal their valuable resources to the public. A case in point is Meta's launch of Llama 3, an open-source large language model that exhibits performance comparable to state-of-the-art proprietary models.
In this research project, we investigate the above-described developments in the field of open-source AI that have brought renewed attention to open-sourcing and selective revealing – phenomena that were initially researched in the context of open-source software over two decades ago. While underlying principles may still apply, we expect crucial mechanisms to be shaped in novel ways by the unique characteristics of AI technology to explain recent observations. Thus, we seek to understand: What motivates technology companies to open-source their AI resources while others choose not to? And how do the unique AI characteristics influence the companies’ revealing strategies?
Contact: Leonard Hanschur
Technology adoption of GenAI in the professional service sector
Integrating Generative Artificial Intelligence (GenAI) into the operations of professional service firms promises a new era of strategic excellence and operational efficiency. Given the significant anticipated potential of GenAI, an increasing number of firms are intensively investing in GenAI solutions for both their clients and internal operations. To fully capitalize on this potential, understanding the human-technology dynamics that influence its adoption is crucial. While GenAI offers revolutionary prospects by enhancing decision-making, automating complex processes, and generating innovative solutions, there is a risk of hesitancy among employees and management, which could slow down the technology's adoption.
This research project explores the influential factors affecting GenAI technology adoption and examines adjacent topics such as the impact on value proposition, effective upskilling approaches, and strategies for organizational enablement.
Contact: Dennis Nesemeier
Patent pools’ growth dynamics
When patent holders collaborate to collectively license their patents, they form what is commonly known as "patent pool". More than two decades after the US Department of Justice’s (USDOJ) legal and antitrust deregulation of patent pools in 1997, these entities are undergoing a process of modernization, improving their licensing arrangements, royalty rate schemes, and commercial behaviors. Especially in the context of Information Communication Technology (ICT), patent pools of Standard Essential Patents (SEPs) have seen a growing recognition of their significance due to their benefits in offering a streamlined mechanism for licensing. To some extent, this fosters higher levels of transparency and predictability within the SEP ecosystem. Nonetheless, economic literature and real-world instances have demonstrated that patent pools often experience commercial failure with a subsequent shutdown of the licensing program or a merger with other pools. Hence, a pressing question arises: under which renewed conditions do patent pools related to ICT standards emerge and successfully compete with their peers? By identifying the incentives for licensors and licensees to join pools through a mixed-methods study, we seek to understand the underlying causes and motivations behind the dynamic emergence of patent pools.
Contact: Pietro Fantini
Patent licensing
Within a few weeks German courts granted four injunctions against a German premium carmaker in 2020, sending shockwaves through the automotive industry. Several patent owners sued for infringement of standard-essential patents (SEPs) on LTE which is leveraged by telematic control units (TCUs), a part of today’s connected cars. Despite the argument that carmakers typically purchase components free of third party rights, thus, leaving full responsibility for licensing with suppliers, some patent owners refused to license suppliers. Based on the licensee's characteristics and level in the value chain, we distinguish two modes of patent licensing: integrated licensing, where the licensee is also the implementer of the patented knowledge (e.g., a carmaker taking a license for engine technology); and bifurcated licensing, where the implementer of the patented knowledge is upstream from and a supplier to the licensee (e.g., a carmaker taking a license for a TCU).
We investigate for SEPs and non-SEPs how common bifurcated licensing is, in which industries we observe it, and what its determinants are by applying a mixed-methods approach. Our analysis is based on a qualitative interview study and a quantitative study of patent license agreements. We want to derive implications from bifurcated licensing for IP managers and policy makers and contribute theoretically to research on value capture.
Contact: Adrian Göttfried