Yonatan Belinkov
CS Taub Building 733
Technion
Haifa 3200003, Israel
I’m an associate professor at the Technion Taub Faculty of Computer Science with a secondary affiliation at the Faculty of Electrical and Computer Engineering. I am a former Azrieli Faculty Fellow.
I work on Artificial Intelligence and Machine Learning, especially large language models, multi-modal models, and other natural language processing models. Main research interests:
- Interpretability, robustness, safety, and controllability
- AI for science, especially biological language models
- Emergent and multi-agent communication
I spent the 2025-2026 academic year at the Kempner Institute.
I was previously a postdoc at Harvard SEAS, where I was affiliated with the Mind, Brain, Behavior initiative and worked mainly with Stuart Shieber. I was also affiliated with the NLP group and the CCNLab at Harvard.
During my postdoc I also worked with the Spoken Language Systems group at MIT CSAIL, where I completed my PhD under the supervision of James Glass. My PhD thesis analyzed internal language representations in deep learning models, with particular applications to machine translation and speech recognition.
Before coming to MIT I worked as a software engineer in IntuView while pursuing a Master’s in Arabic Studies at Tel Aviv University with a thesis on the Arabic dialect of Jisir izZarga. My undergrauate degree was in Mathematics and Arabic Studies, also at Tel Aviv University.
Note to prospective students: Please refer to our lab’s website for information on opportunities and how to apply.
News
| May 2026 | Elected to the Israel Young Academy. |
| Apr 2026 | Featured on Ynet’s 40 Future Young Voices list. |
Selected Publications
- Measuring Chain of Thought Faithfulness by Unlearning Reasoning Steps, EMNLP 2025
- Arithmetic Without Algorithms: Language Models Solve Math With a Bag of Heuristics, ICLR 2025
- Diffusion Lens: Interpreting Text Encoders in Text-to-Image Pipelines, ACL 2024
- BLIND: Bias Removal With No Demographics, ACL 2023
- Locating and Editing Factual Associations in GPT, NeurIPS 2022
- Investigating Gender Bias in Language Models Using Causal Mediation Analysis, NeurIPS 2020
- Don't Take the Premise for Granted: Mitigating Artifacts in Natural Language Inference, ACL 2019
- Synthetic and Natural Noise Both Break Neural Machine Translation, ICLR 2018
- Analyzing Hidden Representations in End-to-End Automatic Speech Recognition Systems, NIPS 2017
- What do Neural Machine Translation Models Learn about Morphology?, ACL 2017