In 2023, AI large models will show strong commercial value and development potential, and science and technology institutions will rush in and implement large model applications. There is a view that the large model is the "battleground" of the head institution, which will determine the scientific and technological level of an institution in the next 5-10 years.
Computing power, algorithms, and data constitute the "troika" in the field of large models, and data, as the core production factor, has become the key point related to the quality and commercialization of large models. As a data-intensive industry, the banking industry has become an important industry that is the first to be explored by large models.
Large models have broad prospects for application in the banking industry
The digital transformation of the banking industry requires the support of large models. Gao Feng, chief information officer of the China Banking (601988) Association, said.
According to analysts, the essence of digital transformation of banks is to "embrace" data and algorithms, and use digital technology to digitally reshape business processes, operation methods, and service models, so as to improve the quality and efficiency of operations and service levels. In general, in the process of digital transformation of banks, "using data" is very important, and artificial intelligence, represented by large models, is an advanced advancement of "using data" and "empowering intelligence", which will empower marketing, operation, risk control, decision-making and other business links more efficiently and deeply, and provide a new paradigm and new momentum for the digital transformation of the banking industry.
Zhang Bin, chief information officer of China Minsheng Bank (600016), pointed out that large language models have a wide range of applications in the financial field and can help improve work efficiency. For example, helping individuals become super-producers can also enhance the experience, such as supporting multiple rounds of conversations - machines interact in a human-like way.
In specific application scenarios, such as the combination of intelligent customer service and digital human, multiple rounds of highly anthropomorphic questions and answers can be answeredIn the field of intelligent investment research, its powerful analysis and refining capabilities and generation capabilities can be usedIn the field of program development, it can assist in coding, testing, and completion;In the field of marketing, it helps precision marketing, including personalized content generation;In the operational domain, it can assist in human interaction, summary and recommendation generation;In the field of risk control, it can realize intelligent risk identification and legal complianceIn terms of knowledge management, it can realize automatic knowledge extraction, knowledge update and maintenance, and provide a better knowledge Q&A experience.
Zhao Huanfang, deputy general manager of the R&D Center of the Agricultural Bank of China, said that domestic financial institutions have given priority to exploring and trying in the fields of intelligent customer service, knowledge question and answer, assisted programming, and smart office.
When it comes to specific applications, financial institutions adopt different strategies based on resource investment, technology accumulation and other considerations. Large banks prefer the self-built model of "deep customization, combination of construction and application", and pay attention to the construction of large model capability system, including computing power, AI platform, model training capabilities, etcIt will combine its own data advantages to train customized large models, focus more on the precipitation of its own large model capabilities, and create a platform-based support model. Small and medium-sized banks may adopt a "direct introduction, lightweight and fast" reference strategy, pay more attention to the direct introduction of industry models and general models, and quickly and lightly connect with scenario applications through API calls and other forms.
The large bank model is still in the initial stage of exploration
Industry insiders believe that the emergence of large models will accelerate the efficiency and quality of banks' digital transformation, and is a financial technology innovation track that banks need to pay attention to, but the application of large models in the banking field is still in the initial stage of exploration. It still faces challenges in many aspects, such as data security, computing power assurance, model construction, open ecology, and ethical risks.
Gao Xulei, director of the Fintech Office of China Merchants Bank, proposed that we should be vigilant against the risks that may be brought about by technological development. Taking large models as an example, the biggest problem with this technology is the "model illusion", that is, everyone is talking about "serious nonsense", but we must not waste food because of choking, we must allow large models to make mistakes like people, and the key is to manage them accordingly. In addition, there may be issues of ethics, copyright, big data, the generation of harmful information, trade secrets, etc.
Zhang Xiaoyan, deputy dean and chair professor of finance at PBC School of Finance, Tsinghua University, said at the "Digital Transformation Prospect Annual Event of the 19th (2023) Digital Finance Joint Publicity Year" that the implementation of large language models in the banking industry also faces many risks and challenges, which requires the industry to be highly vigilant and pay attention to it. For example, large language models may lead to the leakage of confidential bank data and customer privacyThe application of large language models in the banking industry also faces regulatory risks, and the regulatory attitudes of different countries towards large language models are currently quite different. There is an argument that artificial intelligence, including the application of large language models, could become the next flashpoint for financial systemic risk.
Qian Bin, Vice President and Chief Information Officer of Bank of Communications, suggested that attention should be paid to the ethical construction of artificial intelligence and the construction of credible, safe and fair financial applications of artificial intelligence. He pointed out that with the comprehensive application of generative AI in the financial field, the security, fairness and transparency issues that may be caused will receive more and more attention from regulators and market players, and it is necessary to ensure the safety and controllability of the application process and the effective protection of the legitimate rights and interests of service users through effective governance. Financial institutions should pay attention to the accuracy, reliability, and stability of the content generated by large models, and establish effective management and control mechanisms and emergency strategies to prevent problems such as value deviation, algorithm bias, and discriminatory content generation. We should make good use of financial technology responsibly, and strive to practice the value concept of "responsible finance" in the torrent of digital transformation, so that finance can take root for the people and technology for good and roll forward.
A number of industry insiders told China E-banking Network that the future of the banking industry model is promising, and they hope that the industrial chain institutions will cultivate the innovation ecology of the large model, strengthen the implementation of cases and technical exchanges, and share the development dividend of the large model.
At present, the banking industry is thoroughly implementing the spirit of the first-class financial work conference, benchmarking digital financial articles, creating new growth points of artificial intelligence, and providing customers with convenient, safe and warm financial services.
Linghexun plus members, watch more exclusive content for free: 8 major financial columns, exclusive interpretation of the latest and hottest information dry goods, and grasp the market investment trend one step faster.