Jonathan Berant

Jonathan Berant is an Associate Professor at the Blavatnik School of Computer Science, Tel-Aviv University, and a Research Scientist at the Allen Institute for Artificial Intelligence. Currently on leave of absence at Google (based in Seattle) until summer 2026, he has established himself as a leading researcher in Natural Language Processing with significant contributions to semantic parsing, question answering, and weak supervision techniques.

His research focuses on Natural Language Understanding problems including Semantic Parsing, Question Answering, Paraphrasing, Reading Comprehension, and Textual Entailment, with particular interest in learning from weak supervision that is easy to obtain and grounded in the world, as well as tasks requiring multi-step inference or handling of language compositionality. His work bridges theoretical foundations with practical applications in language understanding systems.

Berant's recent publications demonstrate strong trends in language model alignment, robustness, and reasoning capabilities, with significant contributions to understanding how language models process information and how to make them more reliable. His work spans from fundamental NLP tasks to addressing critical challenges in modern large language models.

• Senior area chair outstanding paper award (NAACL 2025)
• Outstanding paper award (ICLR 2024)
• Spotlight talk (NeurIPS 2023)
• Spotlight talk (3% of submissions) (NeurIPS 2022)
• Oral presentation (NeurIPS 2021)

Berant has advised numerous PhD and Master's students who have gone on to positions at leading AI companies and research institutions including Google, AI21, Allen Institute for AI, and academic programs. His teaching includes advanced NLP courses and research seminars at Tel-Aviv University. His educational background includes a Ph.D. from Tel-Aviv University (2006-12) with advisors Ido Dagan, Jacob Goldberger, and Eytan Ruppin, followed by postdoctoral work at Stanford University with Percy Liang and Chris Manning, and at Google Mountain View.