Anca Dragan

Anca Dragan is an Associate Professor in the EECS Department at UC Berkeley and currently leads AI Safety and Alignment at Google DeepMind. She specializes in human-AI interaction, focusing on aligning AI systems with human goals through algorithmic methods. Her work spans robotics, machine learning, and cognitive science, emphasizing safety, collaboration, and understanding human preferences.

• Affiliations: UC Berkeley EECS, Center for Human-Compatible AI, Google DeepMind
• Education: B.Sc. (Jacobs University Bremen), Ph.D. (Carnegie Mellon University)

Research: Her lab (InterACT) develops algorithms for robots and AI to collaborate safely with humans. Key areas include reward learning from diverse feedback (e.g., demonstrations, corrections), game-theoretic coordination, and mitigating risks in advanced AI systems. Notable contributions include Inverse Reward Design and cooperative inverse reinforcement learning.

Publications: Over 100+ papers in top venues (ICRA, CoRL, NeurIPS) on topics like human-robot collaboration, safety-critical systems, and alignment challenges. Recent work addresses risks in large models and scalable alignment techniques.

Awards: Sloan Fellowship, MIT TR35, Okawa Award, NSF CAREER, PECASE. Recognized for teaching excellence (McEntyre Award) and contributions to AI ethics.

Grants & Leadership: NSF and ONR grants, co-PI of Human-Compatible AI. Advised Waymo on safety for autonomous systems. Active in professional roles (e.g., conference chairs, steering committees).

Labs/Teams: InterACT Lab (UC Berkeley), leading safety teams at DeepMind. Collaborations with BAIR Lab and academic-industry partnerships.