The Third Session of the "Pooling Wisdom, Integrating Data: Sharpening the Tools First" Training Activity Was Successfully Held

On the afternoon of March 26, the third session of the "Pooling Wisdom, Integrating Data: Sharpening the Tools First" series of training activities - the special event on "Applications of DeepSeek and Large - scale Models in Economic and Financial Teaching and Research" - was successfully held in Room 1632 of Houzhu Building. This event was organized by the Experimental Teaching Center of the Business School of Beijing Normal University. Li Xinping, the Executive Director of the Financial Engineering Laboratory of Peking University, was invited as the keynote speaker. Sun Yunchuan, the Director of the Experimental Teaching Center of the School, presided over the event, and more than 10 teachers, including Jiang Jie, the Associate Dean of the School, and Hu Haifeng, the Director of the Department of Finance, participated in the activity.

Director Li Xinping first introduced the Financial Engineering Laboratory of Peking University from four aspects: its development history, personnel composition, curriculum teaching, and scientific research. As a leading position in the research of quantitative finance, this laboratory deeply explores the operation laws of financial markets by means of technologies such as mathematical modeling, statistical analysis, computational finance, big data, and machine learning. With the help of cutting - edge technologies, the laboratory has developed a series of platforms and applications, integrating the functions of financial AI teaching assistants and financial AI research assistants.

Subsequently, with his profound professional knowledge, Director Li systematically introduced the applications of large - scale models in multiple fields such as model deployment and education and training, centering around the development history and core technologies of DeepSeek. He pointed out that the application of thought chains and the mixture of experts model has significantly enhanced large - scale models in aspects such as deductive and inductive reasoning, chained - thinking reasoning, and presenting the thinking process. However, in the practical applications of economic and financial disciplines, large - scale models still face problems such as hallucinations in generation and poor data - processing capabilities. To give teachers a more intuitive understanding of these problems, Director Li conducted an on - site demonstration. The result of the demonstration was that the large - scale model generated a pseudo - proposition paper, clearly verifying the existence of the hallucination problem in large - scale models.

Finally, Director Li introduced in detail the functions that financial AI teaching assistants and AI research assistants can achieve. On the one hand, he demonstrated the whole process of quickly retrieving curriculum teachers, teaching arrangements, and syllabus information using a financial AI teaching assistant through practical examples. On the other hand, he demonstrated the method of quickly generating research project application proposals using a financial AI research assistant through practical examples, allowing the teachers present to truly experience the convenience it brings to teaching and scientific research. At the same time, the participating teachers raised relevant questions to Li Xinping, including how to build an intelligent research - assisting system and the cost of system construction, and they discussed these issues.

Through this event, the participating teachers learned about the technical features of DeepSeek and the development and applications of large - scale models, which enhanced their in - depth understanding of large - scale model tools. It has certain inspiring significance for promoting the in - depth integration of AI technology with economic and financial teaching and research and exploring innovative teaching and research models.

Contributed by: Experimental Teaching Center

Edited by: Shen Yan

Reviewed by: Jiang Jie