Thilo Hagendorff

Dr. Thilo Hagendorff is an Independent Research Group Leader at the University of Stuttgart, specializing in the Ethics of Generative AI within the Stuttgart Research Focus "Interchange Forum for Reflecting on Intelligent Systems" (IRIS3D). Previously, he worked for the Cluster of Excellence "Machine Learning" at the University of Tuebingen and was a visiting scholar at Stanford University and UC San Diego. As a lecturer, he teaches at the Hasso Plattner Institute in Potsdam.

His research focuses on AI ethics, machine behavior in language models, and the intersection of machine learning and psychology. Key areas include:

• Ethics of Generative AI and Large Language Models
• Machine Psychology and Behavioral Analysis of AI Systems
• Deception abilities and safety concerns in LLMs
• Speciesism and bias in AI systems
• AI alignment and political bias

Hagendorff's work has revealed critical insights about how language models can develop deceptive capabilities, the emergence of speciesist biases in AI systems, and the ethical challenges of aligning AI with human values. His research on "Machine Psychology" proposes treating LLMs as participants in psychological experiments to explore their behavior. He has published extensively in top venues including PNAS, Nature Computational Science, and Minds and Machines.

Among his notable scientific contributions are:

• Demonstrating that state-of-the-art LLMs have conceptual understanding of deceptive behavior
• Mapping the ethics of generative AI through a comprehensive scoping review identifying 378 normative issues
• Revealing speciesist biases in AI systems that perpetuate discrimination against animals
• Identifying a "looming replication crisis" in evaluating behavior in language models

Hagendorff actively engages with students and researchers, leading an independent research group at the University of Stuttgart focused on Ethics of Generative AI. His work has received significant media attention, with coverage in MIT Technology Review, Nature, Scientific American, and numerous international outlets. He has also been featured on podcasts including Data Skeptic and Prioritäten, discussing ethical implications of AI and animal ethics.