Exploring the Applications of Artificial Intelligence in the Financial Sector through ChatGPT

Applications of AI in finance through ChatGPT exploration.

Author: Yang Tao, Deputy Director of National Finance and Development Laboratory


Since ChatGPT emerged, this AI chatbot has become one of the hottest topics worldwide. Objective analysis shows that AI is more widely used in organizational operations, service capabilities, and risk management in the financial sector, while still insufficient in addressing other financial needs due to technological and institutional factors. The article points out that although ChatGPT further highlights the application capabilities of AI, it still faces many challenges for the financial industry, making it still unable to bring about significant changes in the industry for a long time.

The dialogue robot ChatGPT developed by the American AI research laboratory OpenAI has attracted widespread attention from all walks of life at home and abroad, and has set off a wave of AI fever. At the same time, the digital transformation of the financial industry has become the general trend in various countries and an important reform direction promoted by regulatory authorities in China. Therefore, starting with ChatGPT to deeply analyze the status, opportunities, and challenges of AI applications in the financial field will help to more accurately achieve technological assistance in high-quality financial development.

01 AI Development Status and ChatGPT’s Position

From a macro perspective, whether it is the top-level design of the “14th Five-Year Plan” and the development plan of the digital economy, or the new version of the financial technology development plan and the guidance on digital transformation in the financial field, AI is regarded as the core driving force, key industry and digital foundation of the digital economy. The rapid development of the digital economy has created a good economic and technological environment for AI; at the same time, AI, as a critical new infrastructure, has also provided new impetus for driving the development of China’s digital economy. Overall, the open and shared infrastructure, focused landing tool processes, and diverse and broad application scenarios provide a good application environment and market space for the vigorous development of AI.

From the perspective of technological development trends, ultra-large-scale pre-training models are undoubtedly the focus and hot spot of current AI technology development, and have experienced a big outbreak and “arms race” in the past two years. Overall, large models show a development trend of multi-modal, multi-technology, multi-capability, and multi-application, and have demonstrated good application effects in both ideal laboratory environments and real environments in vertical industries. In the future, a smart system of large and small models and cloud-edge collaboration will be formed.

At the same time, artificial intelligence has brought tremendous impact and challenges to existing ethical norms and social governance. Therefore, how to achieve effective governance of artificial intelligence has become a focus of attention from all walks of life at home and abroad. It can be seen that breakthrough progress has been made in the governance of artificial intelligence at home and abroad. It has entered the stage of institutionalization and implementation from the ideological level, and the development of trustworthy AI has become a core content.

It should be said that artificial intelligence has become the most important “catalyst” for technological innovation, and natural language processing (NLP) related to ChatGPT is considered the “pearl” on the crown of artificial intelligence. We see that the history of the development of artificial intelligence is actually a history of continuously improving model dimensions, from artificial experts writing rules, to machines writing a small number of rules, to machines writing a large number of rules, and finally to transfer learning large models. In this process, ChatGPT uses text learning methods to expand its field, and GPT-3 has 500 billion words and 175 billion parameters. Finally, with the support of massive information, it has achieved a comprehensive improvement in functionality, but also faces challenges such as credible content, data security, and high implementation costs.

02 Opportunities for the Application of Artificial Intelligence from the Perspective of Financial Demands

With the deepening of the construction of the digital economy and digital society, a large amount of data has been generated, providing a broad “soil” for the modeling, training, and application of artificial intelligence. Especially in the financial field, it has accumulated large-scale and high-quality data, and at the same time has multi-dimensional and diversified application scenarios, providing a good opportunity for the vigorous development of artificial intelligence applications. Through the in-depth integration of artificial intelligence and financial sector customer service, product innovation, operation management, and risk control and other business scenarios, the entire process of financial services is reshaped and intelligently empowered to promote financial product innovation, process reengineering, channel integration, and service upgrades, expand the breadth and depth of financial services, and become an important source and driving force for the digital transformation of the financial industry.

Ultimately, the application value of artificial intelligence lies in solving the problems existing in the financial industry, which needs to be analyzed from the perspective of financial demands. Specifically, from the macro and micro levels of the financial industry, the dilemma faced is strategic issues. Facing the increasingly complex economic and financial situation, the strategic formulation of financial institutions has become particularly important. This is not only the “top leader project” of the institution, but also requires the effective combination of vision, logic, and experience, and timely and effective dynamic optimization. The application of artificial intelligence in strategic formulation is precisely perception, reasoning, decision-making, which is naturally possible to be combined with the comprehensive or special strategic formulation of financial institutions, and dynamically and randomly optimized.

Second, structural issues. Although the comprehensive strength of China’s financial industry continues to increase, there are still many structural contradictions that are unbalanced and insufficient in development, which also puts forward requirements for the “short board” of artificial intelligence. For example, if artificial intelligence is applied to the field of wealth management, can it bring changes to the family asset structure and financial asset layout imbalance, which directly affects the major goal of financial assistance to common prosperity.

Third, factor problems. The sustainable development of financial institutions and digital transformation both need to consider the economy, scale, and efficiency of factor inputs, among which the most core are data and people. On the one hand, data has become an important factor of production and a fundamental strategic resource of the country. How to improve the “collection, storage, calculation, management, and use” of data throughout its life cycle in the financial industry, and promote the transformation of data elements into data assets, is the urgent challenge facing the industry, and combining artificial intelligence with big data will stimulate more vitality. On the other hand, financial technology talents are also scarce resources, and artificial intelligence can become an “intelligent assistant” to enhance employee capabilities, or build “digital people” to compensate for team capabilities.

Fourth, organizational operation problems. The digital transformation of the financial industry cannot be separated from the guarantee of organizational structure and operational capabilities. In this process, artificial intelligence can be fully utilized to create automated and intelligent operating models, continuously optimize operational processes, innovate operational models, improve operational service quality, and reduce operational costs, thus supporting comprehensive and intelligent financial services.

Fifth, service capability problems. The service capabilities of financial institutions are reflected in diversified products, sufficient market analysis capabilities, marketing and channel capabilities, customer maintenance and value-added service capabilities, etc. Especially in the design of customized intelligent products, accurate marketing through customer holographic portraits, and consistency of online and offline experiences, there have been effective explorations.

Sixth, risk management problems. The macro and micro risks that the financial industry is facing are more complex. If artificial intelligence can be effectively utilized, it can establish an intelligent risk control model on the basis of integrating and analyzing big data, and become an effective way to identify, monitor, and control risks. On the one hand, build a customer, business, and risk view to dynamically and comprehensively reflect the overall picture of risks; on the other hand, optimize intelligent credit risk assessment and realize the transformation of risk control from numerical control to intelligent control.

Issue Seven: Service Efficiency Whether artificial intelligence is efficient in the financial industry depends on two factors: the financial institution itself and the service entity. On the one hand, in the rapid development of the financial industry in recent years, information technology has had a profound impact on the improvement and transformation of the total factor productivity of the financial industry. One of the values of using artificial intelligence is whether it can further improve the operational efficiency of financial institutions and optimize financial indicators. On the other hand, the financial industry has many responsibilities in promoting inclusive, green, technological, and shared prosperity. How to consider the value of artificial intelligence application for its functional improvement also needs to be considered.

Issue Eight: Collaborative Ecology From open banking to open finance, it has become the mainstream of global innovation. Financial institutions need to share data, algorithms, transactions, processes, and other business functions with the commercial ecosystem to provide services to customers, employees, third-party developers, financial technology companies, suppliers, and other partners of the ecosystem, thus creating digital financial ecology with “smart, open, shared, agile, and integrated” as its main features. With the support of artificial intelligence and big data, it may help further improve the external ecology of financial institutions.

Objectively speaking, artificial intelligence is more widely used in organizational operations, service capabilities, and risk management, and is still insufficient in addressing other financial needs due to technological and institutional factors.

03 Challenges of Applying Artificial Intelligence in Finance

ChatGPT further highlights the application ability of artificial intelligence, but for the financial industry, it still faces many challenges, making it unable to bring significant changes to the financial industry for a long time.

First, data governance. The starting point for the digital transformation of the financial industry is to do a good job of data governance, which requires the real improvement of the data governance system, the enhancement of data management capabilities, the strengthening of data quality control, and the improvement of data application capabilities. Artificial intelligence applications also require high-quality massive data, but the data governance of financial institutions is generally in the initial stage, with low-quality data, data islands, and scattered data, making it difficult to provide sufficient data element support for artificial intelligence.

Secondly, standardization of scenarios. Although the application of artificial intelligence in finance reflects characteristics such as personalization and “thousands of faces”, in the long run, the truly vibrant financial technology innovation scenarios are standardized and universal, rather than differentiated cooperation based on traditional outsourcing models, which is also one of the constraints of existing artificial intelligence financial applications.

Thirdly, high cost threshold of technology and solutions. The technical application of artificial intelligence in financial activities and the setting of solution usually have high deployment costs, which are difficult to meet the needs of most small and medium-sized financial institutions. According to research estimates by Guosheng Securities, the cost of training GPT-3 once is about 1.4 million US dollars, and for some larger LLM (large language models), the training cost ranges from 2 million US dollars to 12 million US dollars.

Fourth, transparency and non-explainability. The so-called explainability means that in a process of action perception or decision-making, sufficient and understandable information needs to be obtained to help make decisions. In the field of machine learning, there is usually an unobservable space known as the “black box” between input data and output answers. Only by developing explainable and trustworthy artificial intelligence financial applications can user trust, model auditability and risk reduction be achieved.

Fifth, internal coordination of the organization. For financial institutions to apply cutting-edge technologies such as artificial intelligence, it is usually difficult to form an effective “incentive compatible” mechanism, which promotes internal stakeholders to reach consensus and maximize the value of technological innovation. Therefore, how to optimize the organizational coordination mode through rule design while optimizing the technical scheme itself is also a challenge that artificial intelligence cannot avoid.

Sixth, responsibility sharing. The product design and business operation of financial institutions have certain characteristics and various complex risks. Therefore, based on the logic of controllable risks and financial consumer protection, any financial activity needs a clear responsibility sharing mechanism. After the introduction of artificial intelligence, there may be some new ambiguities in the original balance of rights and responsibilities in the business processes of financial institutions, and further exploration is urgently needed from the aspects of institutional rules, business practices, technology and business, and the relationship between models and people.

Seven is compliance and ethics. With the rapid development of financial technology, regulations in various countries are keeping up with the times. In the face of dynamic regulatory principles and models, there is more prominent compliance pressure on the application of artificial intelligence in finance. At the same time, ethical challenges in financial technology such as algorithmic discrimination, big data price discrimination, and information leakage have also brought “shadows” to the application of artificial intelligence, and deeper exploration is still needed on how to use “responsible” technological innovation to create “warm” financial services.

All in all, the picture of artificial intelligence driving the digital transformation of the financial industry has been opened up. However, this is not smooth sailing and still faces many major challenges, urgently requiring self-optimization and continuous “passing.”