The rise of big financial models: What has it brought and what has changed?

The rise of big financial models: What has it brought and what has changed?


The financial industry is no stranger to “models”. What has changed with the rise of the word “big model”?

In recent years, a series of artificial intelligence and related technologies such as big data, machine learning, natural language processing, computer vision, speech recognition and intelligent recommendations have made great progress, providing application support for financial technology from “thinking” to “doing it” . However, it should also be noted that due to the particularity of the financial field, which has strong demands in terms of risk control, security, and efficiency, the application of large models is generally still in the early exploration stage and still faces considerable challenges.

The reporter found out that there are many practical applications of large models in the financial field. Immediate Consumption released the large retail financial model “Tianjing”, the Industrial and Commercial Bank of China launched a general model for the artificial intelligence financial industry, the Agricultural Bank of China released the large financial AI model ChatABC, and the University of Science and Technology iFlytek released the “iFlytek Spark” smart customer service for insurance business and other scenarios, and Bloomberg launched a large-scale language model BloombergGPT for the financial industry…

“Finance is a highly data-driven market, and many financial institutions have rushed to the large model track in the past year.” “Currently, large models are accelerating penetration in information understanding, content generation, knowledge question and answer, code programming and compliance testing.” In At the “Financial New Quality Productivity Innovation Forum” held recently, guests participating in the meeting said that the financial industry, as the bloodline of the national economy, has become a large model and even broader model due to its large user base, large economic impact, multiple service scenarios, and strong connections with people’s livelihood. Artificial intelligence technology application scenarios. Finance and big models “hand in hand” will provide decision makers with more accurate predictions and provide users with more personalized services. Big models will become an important force in promoting this change.

At the forum, focusing on empowering financial risk control, Jiang Changjun, an academician of the Chinese Academy of Engineering, said, “We need to design an online transaction risk control system. ‘Run fast’ means that the system can identify whether behaviors are compliant online in a very short time and in real time.” , ‘Catching accurately’ means being able to locate and prevent problems in time if they arise. The user’s behavior can be condensed through a large model, and the unchanging behavioral texture can be obtained from the changing behavior data.”

“It’s not technology for technology’s sake, but focusing on sustainable development on the business side and management side.” Zeng Gang, deputy director of the National Finance and Development Laboratory, believes that the industry is facing difficulties in acquiring assets and difficulty in reaching customers with traditional services; asset quality problems Changes, some excess industries are undergoing adjustments, which requires increasing risk prevention and control; interest rate spreads have narrowed, and capital replenishment needs attention. The development of large models and artificial intelligence will disintermediate, not only “information intermediaries” but also “knowledge intermediaries” to improve efficiency, adapt to changes in customer needs, and expand new service spaces and scenarios.

“Since August 2023, we have launched the large financial model ‘Sky Mirror’, which has increased knowledge output efficiency by 150%. In cooperation with a bank in Chongqing, intelligent marketing driven by large models can reduce labor costs by more than 80%. The production capacity is more than 6 times that of traditional manual production capacity.” Jiang Ning, CTO of Immediate Consumption, said: “In order to provide solutions in terms of group intelligence and security controllability, personalization and privacy protection, critical tasks and dynamic adaptability standards, we will We will continue to explore and develop key technologies such as model security and controllability, combined AI, continuous learning, and platform service capabilities MaaS.”

Finance is the most standardized and rigorous industry. Some people in the industry said that the expectations for large models may be explored in those business links that have not been solved for a long time, and extended to the core financial business, which may bring greater surprises.


Source link