Guest Speaker: Chen Jingyuan

        Chen Jingyuan is a "Hundred Talents Program" researcher and doctoral supervisor at the School of Education, Zhejiang University. His research focuses on educational data analysis and mining, as well as the understanding and generation of multimodal educational resources. He has led several key projects, including the ​Zhejiang Provincial Leading Goose Science and Technology Plan Project, the ​National Natural Science Foundation Youth Project, and the ​Shanghai Science and Technology Plan Project. Additionally, he serves as the sub-project leader for major initiatives such as the ​National Natural Science Foundation Major Project, the ​Ministry of Science and Technology's Science and Technology Innovation 2030—New Generation Artificial Intelligence Major Project, and the ​Young Scientist Project. He has been awarded the ​Zhejiang Provincial Science and Technology Progress Award (First Prize) and the ​IET Smart Cities Best Paper Award, and was recognized as one of the ​Top Ten Outstanding Youths in Yuhang District, Hangzhou (2022) and the ​Shanghai Youth Science and Technology Rising Star (Class A, 2022).

        Abstract:

        This report employs analogies to demystify language understanding, contrasting the human "guessing game" with AI's "probability maze." Both humans and AI capture semantics through contextual puzzles: humans rely on experiential knowledge to fill gaps, akin to the "idiom chain," while large language models (LLMs) use trillions of parameters as "memory dictionaries," generating responses via pattern matching. Though their approaches differ, they form a fascinating ​cognitive mirror of shared logic.

Chinese website redirect link:https://www.bj-first.org.cn/single?id=334