Hello! I am a Senior Research Scientist at Google DeepMind and a Visiting Lecturer at Harvard University. My research is driven by the pursuit of intelligent decision-making, with a current emphasis on understanding and improving the alignment between human judgments and Bayesian beliefs of AI systems. Examples:
Understanding AI models via
Improving AI’s belief alignment to humans by
- pre-training Gaussian processes to align with expert beliefs,
- proactively gathering information from humans in image generation and reasoning tasks.
My prior experiences are in machine learning and robot learning, including Bayesian optimization, learning for task and motion planning, active learning, and Gaussian processes. I completed my Ph.D. in Computer Science at MIT, advised by Leslie Kaelbling and Tomás Lozano-Pérez. I received my S.M. in Electrical Engineering and Computer Science from MIT, advised by Stefanie Jegelka and Leslie Kaelbling. Click here for my CV.
Bulletin
- Our paper QuestBench: Can LLMs ask the right question to acquire information in reasoning tasks? was accepted at NeurIPS 2025!
- Checkout 2025 Seminar Series on Bayesian Decision-making and Uncertainty.
- Our paper Proactive Agents for Multi-Turn Text-to-Image Generation Under Uncertainty was accepted at ICML 2025!
Contact
wangzi ‘at’ google ‘dot’ com
For mentorship and career opportunities, please visit Google CSRMP, GDM Scholarships and GDM Careers.
ziw ‘at’ csail ‘dot’ mit ‘dot’ edu
Due to bandwidth limitations and occasional spam filter issues, I may not be able to respond to every email. Thank you for your understanding.