Sabine Wiesmüller

Title of Dissertation:
Forms and Interactions of AI Governance

Supervisor: Prof. Dr. Josef Wieland
University: Zeppelin University Friedrichshafen
Scholarship: KSG Scholarship
Cohort: 3. Cohort, 2016-2020
Email: Diese E-Mail-Adresse ist vor Spambots geschützt! Zur Anzeige muss JavaScript eingeschaltet sein!

• Short Abstract

  The development and implementation of Artificial Intelligence is one of the main drivers of economic progress. However, through the rise in levels of connectedness around the world, competition rises, too. Caused by the rapid pace of AI development that is driven by global competition, companies find themselves under high pressure to come forward with new innovations - beating their competitors around the globe. Therefore, the thesis aims to reply to the rising demand for stronger regulation asked for by society (Brundage et al, 2018; Bryson, 2018; Cihon et al, 2019). Further, it addresses the already identified gap in academia and practice, requiring the operationalisation and implementation of AI governance, instead of the mere creation of additional ethical guidelines for AI development and implementation (Hagendorff, 2020). Thus, the author aims to provide a structural approach to enable the positive outcomes of AI development and research, while at the same time identifying the negative externalities of this progress to manage them accordingly, by applying AI governance to firms.

  Thus, the fierce competition in the private sector, on the one hand, and the demand for AI governance on the other hand, seemingly oppose one another. Additionally, the negative externalities perceivable in practice, which affect society in an unchecked manner, seem to demand a collaborative approach to prevent these effects from happening. Hence, the thesis aims to move from a problem-oriented perspective of classic AI ethics research to a rather solution-oriented approach in AI governance.

  In this context, the thesis focuses on the governance of AI and the demand on the private sector to take regulatory and governance measures. The firm as perspective and governance structure is chosen for two main reasons: First, corporations are understood to be the main driver for the AI revolution, and second, public sector regulation currently fails to address issues of AI governance. Based on a stakeholder approach, this thesis takes the view that corporations do need to internalize the negative externalities coming with corporate decision-making. Thus, applied to the AI context, firms need to acknowledge and address the risks coming with research on AI and the implementation of AI-based products and processes. While Wieland (2018, 2020) applies this view to the phenomenon of globalisation, from the author’s point of view, this concept is equally applicable to AI governance. ‘Contract/private ordering/governance leads naturally into the reconceptualization of the firm not as a production function in the science of choice tradition, but instead as a governance structure’ (Williamson, 2002, p. 191). Therefore, by focusing on private sector ordering and allowing corporations to integrate other system logics than the one of the economic system, they can engage, e.g., as actors of civil society, and thus, realize private sector governance.

  After having established AI research and the implementation of these technologies as a phenomenon of trans-sectoral, global and cross-industry relevance, the thesis proceeds to present a first theoretical conceptualisation of AI governance. Again, to address the correlating ethical risks and the rising number of societal concerns about societal shifts and consequential inequalities, the author focuses on developing a structural model for AI governance. The model procedurally integrates an ethical dimension, without, however, imposing one singular normative position. To achieve a first conceptualisation of AI governance, the approach is presented in form of a self-developed function and framework for relational AI governance framework. Both self-developed instruments are based on the aforementioned Relational Governance approach (Wieland, 2018, 2020). They aim at summarising existing streams of literature, which are understood to be integral parts to AI governance. Methodologically, the themes within the AI governance framework are derived inductively from existing literature, by a semi-structured review of existing literature that identifies prevalent themes and streams in research. Following the research’s methodology, the thesis proceeds to depict the categories and mechanisms integrated in and inherent to the relational AI governance framework. The function represents all elements inherent to Relational AI Governance and includes the necessary dimensions an organisation, which applies the framework, needs to consider. The framework itself serves to systematically elaborate on the scope of and relevant disciplines for AI governance, and to analyse as well as abstractly present the complex nature of the phenomenon. The summarising function represents the three main disciplines as identified, namely research in AI, AI ethics, and governance measures. The overall content of the framework is depicted by dedicating one dimension to each aforementioned discipline, inherent to the phenomenon of AI. Thereby, the framework can serve other scholars as the necessary base to position their work – as it did for the work at hand – and help to systematically identify apparent streams and gaps in research. Apart from providing patterns to derive theoretical insight, function and framework shall serve to enable a better understanding of the themes and questions practical AI governance needs to address.

  By offering the first conceptualisation of AI governance, the framework contributes to further structuring this comparatively young field of research. For instance, within the economic dimension, looking into the global race dynamics and fierce competition in AI research will help contextualise possible approaches and give a more in-depth understanding of the motivation of and restrictions for companies when planning to engage in AI governance measures. The practical economic and societal value of this thesis is constituted by the fact that solving this challenge seems to be crucial when developing AI governance measures, which companies shall be able to apply and implement effective AI governance without fearing immediate competitive disadvantages.

• PhD Related Publications

  • Sabine Wiesmüller. The Relational Governance of Artificial Intelligence: Forms and Interactions. Springer, 2023. https://link.springer.com/book/10.1007/978-3-031-25023-1