Meet Prof. Yonatan Belinkov

The Technion is home to many brilliant researchers responsible for important breakthroughs in core AI technologies. Assoc. Prof. Yonatan Belinkov of the Henry and Marilyn Taub Faculty of Computer Science is one of them. An expert in machine

learning, and especially in large language models (LLMs), he is currently on sabbatical at the Kempner Institute for the Study of Natural and Artificial Intelligence at Harvard University, where he is researching the science of AI, including interpretability and controllability of AI models. In a nutshell, he is trying to understand how systems such as Gemini and ChatGPT make decisions. “Interpretability research involves opening their ‘black box.’ The goal is to control these AI models and make them more robust, efficient, and reliable,” he explains.

Prof. Belinkov focuses on studying ‘hallucinations’—when an AI platform replies to a question with nonsense. “We want to know how to detect these hallucinations in advance and intervene.” He also develops ways to erase knowledge from AI platforms for the sake of privacy or security. “For example, how can I delete information that I don’t want ChatGPT to know about me? Or how can we automatically delete instructions for building a bomb, for example?” he elaborates. Prof. Belinkov’s recent research integrates AI and biology, aiming to promote drug discovery, material design, and more.

Belinkov is proud of the high level of AI research conducted at the Technion and believes that the university is well-positioned to lead the AI revolution. “The Technion has many world-class experts in both AI and sciences and is a great place for multidisciplinary research that combines AI and other fields, such as chemistry, biology, materials science, and others. There is a large amount of data available at the Technion, which is a big advantage.”

This article originally appeared in the Technion President’s Report 2026. The full report is available here.