Time: June 15th (Thursday) 10:00 am -11:30 am

Location: 240 Lecture hall of the main building

Report Title: The Impact of ChatGPT and Big Models on the Paradigm of Economic Research

Summary of report content:

The rapid development of AI technology represented by ChatGPT, especially the invention and application of Big data model, is profoundly changing the way of production, lifestyle and governance of human society, as well as the research paradigm, research methods and research tools of economics. This report will explore the impact of large models on economic research paradigms. For a long time, economics and econometrics have generally aimed at a minimalist model with economic interpretability, interpreting typical empirical characteristics, exploring economic operating laws, and making economic predictions based on the minimalist model. Although Big data and AI have improved the forecasting ability of the simple model to a certain extent, due to the complexity and time-varying nature of the economic system, economic forecasting still faces great challenges. In the era of digital economy, massive Internet data has emerged. Large models based on Big data and AI can ensure that the sample size for training is far greater than the number of model parameters, thus effectively avoiding the "Curse of dimensionality". Therefore, how to expand the scale of models on the basis of big data and AI is a new research paradigm worth paying attention to in economics, such as how big models will have a profound impact on computational economics, econometrics, and the hypothesis of rational economic people. At the same time, this report will also discuss the limitations of AI technology. For example, AI Causal inference is a statistical relationship inference in nature, and AI has not changed the nature of economic empirical research, that is, economic empirical research based on Big data still infers the population from samples and makes predictions outside samples. In addition, due to the different operational characteristics of the economy at different historical stages, the laws of economic operation reflect significant temporal variability, which makes AI's prediction of the economy more challenging.

Reported by:

Hong Yongmiao, Guan Zhaozhi, Principal investigator of the Institute of Mathematics and Systems science of the Chinese Academy of Sciences, Dean of the School of Economics and Management of the University of the Chinese Academy of Sciences, Academician of the Academy of Sciences for Developing Countries, Member of the World Econometric Society, project leader of NSFC's Basic Science Center for "Econometric Modeling and Economic Policy Research", and Vice Chairman of the Steering Committee for Economics Teaching in Universities of the Ministry of Education. Formerly served as a Chair Professor in Economics and International Studies at Cornell University in the United States (2010-2020), and President of the Chinese Economic Society for Studying in the United States (2009-2010).

Research fields include econometrics, time series analysis, financial Metrology, statistics, and China's economy. In Annals of Statistics, Biometrika, Economics, Journal of American Statistical Association, Journal of Political Economy, Journal of Royal Statistical Society B, Quarterly Journal of Economics, Review of Economic Studies, Review of Financial Studies More than 120 articles have been published in mainstream Chinese and English journals in economics, finance, and statistics, including Economic Research and Managing the World. Published Probability theory and Statistics, Advanced Econometrics, Foundations of Modern Economics: A Unified Approach and other Chinese and English works. From 2014 to 2022, he has been selected as one of the highly cited scholars in Elsevier Economics/Statistics China for 9 consecutive years.

(Undertaken by: Department of International Trade and Finance, Research and Academic Center)