Digital Twins Lab

The Digital Twins Lab explores digital twin technology and its real-world use cases with a commitment to transparency, independence, interdisciplinary collaboration, rigor, and accessibility. We develop and evaluate methods, datasets, and tools that help researchers and practitioners understand when digital twins work, when they fail, and how they can be deployed responsibly.

Digital twin technology leverages large language models to simulate human behavior and complex systems, enabling rapid testing, experimentation, and iteration. It has the potential to transform marketing, social science, behavioral research, and business operations, among many other domains. More broadly, digital twins enable high-precision personalization, efficient synthetic data generation, and scalable reinforcement-learning environments—forming a foundation for world models and effective human–AI collaboration.

Our research spans measurement and evaluation, human–AI alignment, domain-specific applications, and reproducible research practices. We emphasize impact beyond the lab by producing open resources, clear reporting standards, and public-facing artifacts that support broad adoption, critical scrutiny, and responsible use.

We are hiring predocs and postdocs. For more information, please contact aib@gsb.columbia.edu.