Subjects: Other Disciplines >> Synthetic discipline submitted time 2023-03-28 Cooperative journals: 《中国科学院院刊》
Abstract： While cross-border data flow considerably improves the effectiveness of cross-border collaborations, it also faces issues of data sovereignty, national security, conflict of interest, privacy protection and data regulation. It has become a link to the global economy and the center of the game of the new order. First of all, this study analyzes the role of cross-border data flow in economic globalization from the perspective of the economic impact of data elements. It then analyzes the constraints of cross-border data flow in the global context and lists the benefits and drawbacks of China’s data element market construction. In the end, based on the practice of China’s digital economy development, the study puts forward policy recommendations of better integration of government with effective market, promoting R&D application of digital privacy technologies, playing the role of early and pilot implementation in free trade zones and free port areas, implementing differentiated regulatory policies of classification and grading, initiating multilateral agreements to export Chinese solutions for international data governance, and adhering to systemic thinking to lead the globalization of digital economy. It aims to promote the safe and orderly flow of cross-border data, so that China can better grasp the strategic opportunities in the globalization of digital economy.
Subjects: Other Disciplines >> Synthetic discipline submitted time 2023-03-28 Cooperative journals: 《中国科学院院刊》
Abstract：以 ChatGPT及其大语言模型为代表的人工智能将对经济学研究范式产生深远影响。目前，经济学实证研究一般使用参数维数较少、经济可解释性较强的小模型。然而，经济金融系统是一个超高维动态复杂系统，受多种因素的影响，且这些因素之间的关系呈现非线性与时变性特征，小模型无法刻画其本质规律。大模型可有效减少系统偏差，更好刻画复杂经济系统的特征与运行规律；而海量数据的使用可避免模型过度拟合，使大模型具有较好的泛化能力即样本外预测能力。为支持经济学及社会科学其他领域的大模型的估计、推断与预测，需要利用人工智能技术整合各种异构、异源、异频数据，构建大规模计量经济学数据库，并加强大算力等信息技术基础设施建设。目前，ChatGPT 及大模型等前沿人工智能技术仍存在局限性，如：无法像人类一样进行批判性思考或想象，只有预测能力；基于大数据的人工智能因果推断本质上是一种统计关系推断，需要引入经济理论或实验方法帮助识别真正的因果关系；人工智能技术不能改变经济学实证研究从样本推断总体性质的本质；同时，由于互联网大数据存在大量虚假信息或样本选择偏差等问题，基于人工智能所获得的结论的可靠性需要验证。
Hits
Hits
Downloads
Comment