creators

Jacob Steinhardt - Berkeley Professor, Transluce CEO

Jacob Steinhardt teaches statistics and EECS at UC Berkeley and runs Transluce, the nonprofit lab he co-founded in 2024 to understand frontier AI systems.

Jacob Steinhardt is an Associate Professor of Statistics and of Electrical Engineering and Computer Sciences at UC Berkeley, where he is part of BAIR and CLIMB. He took his PhD in computer science at Stanford under Percy Liang after undergraduate study at MIT, and worked at OpenAI and as a fellow at the Center for Security and Emerging Technology. His research is on making machine learning systems understood by and aligned with people: reward specification, robustness, and reading the latent activations of neural networks rather than only their outputs. He co-wrote "Concrete Problems in AI Safety", the 2016 paper that set out much of the technical AI safety agenda, and was a senior contributor to MMLU, one of the most used benchmarks for language-model capabilities. In October 2024 he co-founded Transluce with Sarah Schwettmann, a nonprofit lab building open tools for understanding and steering frontier AI systems; he is its chief executive. He is also a scientific advisor to LawZero, the nonprofit AI safety lab whose scientific director is Yoshua Bengio.

Specialization

AI alignment, AI safety, interpretability, model evaluation

Country

United States