Essays on Economic Decision-Making in the Era of Machine Intelligence

by Alexander Erlei
Doctoral thesis
Date of Examination:2025-03-14
Date of issue:2025-04-09
Advisor:Prof. Dr. Kilan Bizer
Referee:Prof. Dr. Kilan Bizer
Referee:Prof. Dr. Holger Rau
Referee:Prof. Dr. Ujwal Gadiraju
Persistent Address: http://dx.doi.org/10.53846/goediss-11169

 

 

Files in this item

Name:Dissertation_AE.pdf
Size:18.5Mb
Format:PDF

The following license files are associated with this item:

Abstract

English

This dissertation studies how human decision-makers react to and adopt novel technology, with a particular focus on algorithmic prediction systems and artificial intelligence. To that end, it combines theory-guided economic experiments with concepts from computer science. As machines increasingly permeate economic decision-making contexts, they operate in both strategic settings — where game-theoretic concepts become crucial — and individual decision environments where prediction quality is paramount. In strategic contexts, AI agents now participate in market interactions previously reserved for humans, raising important questions about cooperation, market efficiency, and welfare distribution. Of particular interest are expert markets characterized by information asymmetries, where AI prediction systems may significantly impact service quality and resource allocation across domains like medicine, finance, education, and law. In individual decision contexts, despite the superior predictive capabilities of algorithms compared to noisy human judgment, research consistently documents human under-utilization of artificial systems. This dissertation addresses these phenomena across seven chapters: Chapters 2 examines how experts' signaling concerns affect the dissemination of algorithmic decision aids on credence goods markets, while Chapter 3 analyzes the concurrent effects of generative AI on market efficiency and surplus distribution through price setting and consumer choices; Chapter 4 analyzes strategic interactions in AI-assisted bargaining; and Chapters 5-7 investigate fundamental explanations for algorithm aversion, exploring human reliance on prediction algorithms and large language models through rational choice theory and contextual multi-armed bandits. Together, these studies advance our understanding of human-AI interactions in increasingly automated economic environments.
Keywords: experimental economics; artificial intelligence; prediction algorithms; game theory; rational choice; social preferences; computer science; generative AI; credence goods; decision making; human-AI interaction
 
Statistik