Scholar
Florian Mai
Google Scholar ID: MfETM20AAAAJ
Junior Research Group Leader, Uni Bonn
Citations & Impact
Citations
450
H-index
10
i10-index
11
Publications
20
Co-authors
18
list available
Publications
Resume (English only)
Academic Achievements
- Publications: 1. 'Superalignment with Dynamic Human Values' accepted at ICLR 2025 Workshop on Bidirectional Human-AI Alignment; 2. JQL paper accepted at EMNLP 2025; 3. Survey-to-Behavior preprint; 4. Preprint on in-training defenses against emergent misalignment; 5. JQL preprint.
- Academic Activities: 1. Organized the International Conference on Large-Scale AI Risks; 2. Participated in a panel discussion on trustworthy AI at the Deutsches Museum Bonn; 3. Offered a seminar course on the ethics of Artificial General Intelligence.
Research Experience
- Position: Junior Research Group Leader at the mAI alignment lab, University of Bonn. Research Projects: 1. Scalable Oversight by Learning to Decompose Tasks; 2. Emergent Misalignment; 3. Value Alignment.
Background
- Research Interests: Scalable Oversight, Value Alignment, Emergent Misalignment, Reasoning Models, LLM Training. Professional Field: AI Alignment and Safety Issues. Introduction: Leads the mAI alignment lab at the University of Bonn, focusing on ensuring that current and future advanced AI systems act reliably in accordance with human values.
Miscellany
- Personal Interests: Organizes an AI safety reading group, discussing recent papers on alignment and more.
