Vera Schmitt

Founder and Head of XplaiNLP Research Group at Johannes Gutenberg University Mainz and Technical University Berlin.
GI Junior Fellow and member of CERTAIN.
Guest Researcher at DFKI.

Research Interest of the XplaiNLP Research Group: Transparent and Trustworthy AI for Decision Support in High-Stakes Domains

The XplaiNLP group develops Intelligent Decision Support Systems (IDSS) that are explainable, robust, and human-centered. Our research spans the entire IDSS pipeline, combining NLP and (M)LLMs, Explainable AI and mechanistic interpretability, HCI, and legal analysis so that AI-driven decisions align with ethical and societal values. We focus on high-risk applications where human oversight is critical:

• Trustworthy Information Processing: Detecting and verifying mis- and disinformation, propaganda, and foreign information manipulation and interference (FIMI) across text, images, and video, improving factual consistency in generated text, and protecting sensitive data such as patient records and speech.
• Interpretability & Explainability: Making model behaviour transparent through XAI and mechanistic interpretability methods, and designing actionable, user-centered explanations that support informed human-AI decision-making.
• Robust & Compliant LLMs: Benchmarking LLM robustness and European value alignment, building explainable LLM monitoring, and supporting compliance with the AI Act and related regulation.

Through interdisciplinary and international collaboration with partners in academia, media, and industry, XplaiNLP shapes AI that is not only powerful but also transparent, fair, and accountable.

News 📰