Christian Nickel

Christian Nickel

Computer Scientist and Researcher @ RC-Trust. Exploring AI-Reasoning, -Safety, -Security and its intersection with Human Behavior. Passionate about both progress and safety in the agentic age, incorporating multidisciplinary approaches to push the boundaries of AI and its applications.

Publications, Presentations & Projects

Publications & Presentations

Forthcoming

2026

  • Understanding Artificial Theory of Mind: Perturbed Tasks and Reasoning in Large Language Models
    Christian Nickel, Laura Schrewe, Florian Mai, Lucie Flek
    Preprint — February 2026
    [arXiv]

2025

  • Understanding Theory of Mind in Large Language Models: The Role of Perturbations and Chain-of-Thought Prompting
    Christian Nickel, Laura Schrewe, Lucie Flek
    ToM4AI Workshop @ AAAI 2025, Philadelphia, PA, USA — March 3, 2025
    Short paper and poster presentation
    [PDF] · [OpenReview]

    2024

  • Probing the Robustness of Theory of Mind in Large Language Models
    Laura Schrewe, Christian Nickel, Lucie Flek
    WiNLP @ EMNLP 2024, Miami, FL, USA — November 2024
    Short paper and poster · 🏆 Best Poster Award
    [arXiv] · [OpenReview]

Research Projects

Ongoing

  • LLM Agent Steerability in Social Environments (2024 – present)
    In collaboration with Dr. Florian Mai, University of Bonn Investigating the steerability of LLM agents with respect to predefined rules, values, and properties in social environments. We analyze how targeted interventions shape model reasoning and subsequent behavior — evaluating the impact of reasoning chains on resulting behavior and going beyond purely linguistic analysis to study behavioral shifts directly. Preprint forthcoming.
  • Strategic Human–AI Interaction in Economic Games (2025 – present)
    RC-Trust / HUAM Group, University of Duisburg-Essen Studying how human strategic behavior shifts in economic games when participants face AI opponents rather than human ones. The project investigates behavioral patterns, adaptation strategies, and the cognitive effects of perceived AI agency on human decision-making.

Completed

  • Theory of Mind in Large Language Models (2024)
    Data Science & Language Technologies Group (CAISA Lab), Bonn-Aachen International Center for Information Technology (b-it), University of Bonn Developed a comprehensive human-written and annotated benchmark dataset of 68 tasks across 10 complexity classes for evaluating Theory of Mind in LLMs. Created novel metrics for assessing Chain-of-Thought reasoning correctness and investigated robustness of ToM capabilities under various perturbations. Presented at WiNLP @ EMNLP 2024 (Best Poster Award) and ToM4AI Workshop @ AAAI 2025 (PDF). Extended full paper: [arXiv:2602.22072]
    (an early paper focusing on the robustness part: [arXiv])
  • Human vs Machine Attention Analysis (2023)
    Lab: Development and Application of Data Mining, University of Bonn Compared human eye-tracking data with machine attention mechanisms in NLP. Analyzed correlation between human fixation patterns and machine saliency scores, and evaluated differences between English and German language processing.
  • Ethics in Autonomous Weapon Systems (2023)
    AI Ethics Seminar, University of Bonn Analyzed ethical implications and the legal framework of autonomous weapons systems. Evaluated concepts of meaningful human control in military AI applications.
  • Self-Sovereign Identity and DLT Solutions (2022 – 2023)
    Fraunhofer Institute for Applied Information Technology (FIT) Worked on Distributed Ledger Technology (DLT/Blockchain) solutions with a focus on usability and interoperability.
  • Privacy Analysis of IoT Devices (2021)
    Bachelor Thesis, University of Bonn Developed a methodology for analyzing privacy compliance of IoT devices. Conducted comparative analysis of smart cameras from different jurisdictions, implemented a network traffic analysis setup using Raspberry Pi, and created an evaluation framework based on GDPR requirements.
  • RansomDenied: Usable Ransomware-Resistant Backup System (2020)
    Bachelor Lab Project, University of Bonn Designed and implemented a user-friendly backup system resistant to ransomware attacks. Conducted user research including expert interviews and requirements analysis, applying user-centered design principles to a security-critical application.
  • ReselMusic: Cloud-based Multi-User Social Playlist Service (2014)
    Private Project Designed and implemented a cloud service for multimedia aggregation in playlists. PartyMode allowed users to check in via QR code and contribute to a shared playlist in real time.