Enable AI, AI Enables - Toward Autonomous, Economically Acting Machines

Titelangaben

Duda, Sebastian:
Enable AI, AI Enables - Toward Autonomous, Economically Acting Machines.
Bayreuth , 2025 . - II, 70 S.
( Dissertation, 2025 , Universität Bayreuth, Rechts- und Wirtschaftswissenschaftliche Fakultät)

Volltext

	Format: PDF Name: Dissertation.pdf Version: Veröffentlichte Version Verfügbar mit der Lizenz Deutsches Urheberrechtsgesetz. Das Dokument darf zum eigenen Gebrauch kostenfrei genutzt werden. Darüber hinaus bedürfen die Vervielfältigung, Bearbeitung, Verbreitung und jede Art der Verwertung der schriftlichen Zustimmung des jeweiligen Rechteinhabers.
	Download (279kB)

Abstract

The emergence of the machine economy (ME) characterized by autonomous entities capable of economically motivated action and collaborative value creation presents organizations with major challenges and opportunities. Currently, there are pivotal developments toward machine autonomy which is key for the ME, e.g., through advances in artificial intelligence (AI). However, many organizations are still struggling to effectively develop and integrate AI into their operations. A major obstacle to exploit AI's potential is data scarcity, which hinders training, optimization and scaling of AI systems. The overarching aim of this thesis is to guide organizations toward the ME by improving AI development. This is addressed by investigating how to enable organizations to manage AI development resources more efficiently, how to enable organizations to mitigate AI’s data scarcity issue through distributed machine learning (DML), and how to understand the path toward the ME. The presented research is structured as a series of six essays. The first essay conceptualizes AI development through a resource portfolio perspective, examining how investments affect outcomes. The second essay proposes a research agenda focused on the implementation of privacy-enhancing technologies in AI development. The third and fourth essays design, implement and evaluate DML approaches in specific application contexts to overcome data scarcity, i.e., essay three implements a federated prescriptive process monitoring approach whereas essay four proposes a split learning architecture to collaboratively improve demand forecasting in supply chains. The fifth essay constructs a comprehensive five-layer model capturing the functionality and potential of ME entities. Lastly, the sixth essay develops a maturity model of ME entities which draws the path toward the emerging ME. This thesis contributes to research by addressing various research gaps by providing prescriptive and design knowledge for AI development as well as presenting a unified understanding of the ME. In conclusion, the thesis strives to enable AI for organizations and sheds light on how AI enables the emerging machine economy.

Abstract in weiterer Sprache

The emergence of the machine economy (ME) characterized by autonomous entities capable of economically motivated action and collaborative value creation presents organizations with major challenges and opportunities. Currently, there are pivotal developments toward machine autonomy which is key for the ME, e.g., through advances in artificial intelligence (AI). However, many organizations are still struggling to effectively develop and integrate AI into their operations. A major obstacle to exploit AI's potential is data scarcity, which hinders training, optimization and scaling of AI systems. The overarching aim of this thesis is to guide organizations toward the ME by improving AI development. This is addressed by investigating how to enable organizations to manage AI development resources more efficiently, how to enable organizations to mitigate AI’s data scarcity issue through distributed machine learning (DML), and how to understand the path toward the ME. The presented research is structured as a series of six essays. The first essay conceptualizes AI development through a resource portfolio perspective, examining how investments affect outcomes. The second essay proposes a research agenda focused on the implementation of privacy-enhancing technologies in AI development. The third and fourth essays design, implement and evaluate DML approaches in specific application contexts to overcome data scarcity, i.e., essay three implements a federated prescriptive process monitoring approach whereas essay four proposes a split learning architecture to collaboratively improve demand forecasting in supply chains. The fifth essay constructs a comprehensive five-layer model capturing the functionality and potential of ME entities. Lastly, the sixth essay develops a maturity model of ME entities which draws the path toward the emerging ME. This thesis contributes to research by addressing various research gaps by providing prescriptive and design knowledge for AI development as well as presenting a unified understanding of the ME. In conclusion, the thesis strives to enable AI for organizations and sheds light on how AI enables the emerging machine economy.