Data Intelligence Faces Organizational, Technical and Financial Challenges How Technology Platforms Empower Financial Institutions to "Rise to the Challenge"

21st Century Business Herald reporter Chen Zhi reported from Shanghai

With the increasingly widespread application of AI in data analysis, many financial scenarios have benefited greatly.

Huang Xu, director of digitalization at PwC, revealed to reporters that different financial institutions have achieved remarkable results in the application of AI+ data, such as a large joint-stock bank that is applying data intelligence to anti-fraud risk prevention and control. Specifically, on the one hand, this joint-stock bank uses knowledge graph technology to solve the problem of telecom fraud data expansion, so as to identify and query new risk accounts and zombie accounts more efficiently, on the other hand, it establishes a model library according to risk characteristics, calculates the deviation value of the current case and historical scene characteristics in real time, calculates the probability of case involvement in the case, and uses big data and artificial intelligence technology to quickly establish models such as personal timely risk warning rules to analyze and make decisions on risk events in real time.

"At present, this joint-stock bank has established an enterprise-level overall management structure of anti-fraud prevention and control strategies and rule models, and empowers risk mining in telecom fraud business scenarios, completing the digital and intelligent transformation of risk control of personal accounts involved in cases." He told reporters.

In addition, insurance companies have also made great achievements in AI+ data. For example, a large insurance company integrates their basic information, policy information, vehicle information, claim information, insurance type details and other data for customers who hold a property insurance to explore the possibility of adding life insurance to customers within one year after the product takes effect. With the blessing of AI+ data, insurance companies have greatly increased the success rate of relevant customers through targeted marketing, and the per capita production capacity of insurance agents has also increased by about 25%.

It is worth noting that the increasing popularity of AI+ data also tests the comprehensive capabilities of enterprises and institutions in all walks of life.

Recently, Altair (ALTR.Nasdaq), which develops simulation, high-performance computing (HPC) and artificial intelligence (AI) solutions, released an international survey showing that although most enterprises around the world widely adopt and implement enterprise data and AI strategies, its success depends mainly on three aspects: organizational, technical and financial.

On the organizational side, many of the companies surveyed are struggling to fill a large number of job openings in the field of data science, but the talent shortage and the time required to upgrade the AI skills of existing employees are still common problems faced by enterprises to adopt AI+ data strategies. In terms of technology, more than half of the surveyed companies said that they face many technical limitations, which greatly hinder the development of AI+ data initiatives; In terms of financial issues, although the surveyed companies want to expand their AI+ data strategies, teams and technicians face layers of financial obstacles.

A domestic bank AI risk control personnel also told reporters that in order to improve the identification ability of financial fraud teams, on the one hand, they need to introduce more structured abnormal information (including the combination of static data, transaction data and time series), on the other hand, they need to build AI models to deeply mine the transaction link, transaction network and fraud team structure, but this requires banks to provide a lot of financial support and recruit more talents proficient in AI. "Insufficient awareness" of the long-term resource investment required for AI+ data.

"At present, while we are introducing third-party AI technology institutions for resource empowerment, we are also asking banks to increase investment in relevant AI talent recruitment through internal communication to solve the bottleneck and friction of current AI+ data." He told reporters.

In the view of industry insiders, although the AI+ data model still faces three major challenges in organizational, technical and financial aspects, global data science and machine learning are showing three important development trends.

First, the edge artificial intelligence market will have explosive growth, that is, under the resonance of the acceleration of digital transformation practices of enterprises in various industries such as finance and the Internet of Everything, edge devices and the amount of data generated by them surge, but because a large amount of edge side data is transferred to a central public device for processing is not realistic, so more and more enterprises will choose to analyze directly on the edge side near the data generation, greatly reduce energy consumption through local analysis, and eliminate the privacy issues involved in offloading data to remote computer systems. In the future, edge devices will become smarter.

Second, data analysis will drive business value and make decisions more accurate. That is, decision-centric data and analytics will gradually replace data-analysis-driven decision-making. This requires AI personnel in data analysis to "enter" the business, so that business personnel can have data analysis capabilities and thinking, and make decision-making and data in the work process.

Third, more and more companies will use adaptive AI systems to cope with changes in internal and external factors. As the need for real-time data processing, streaming, and sharing increases, driving the shift of more enterprises to data-analytics-driven enterprises. As a result, companies need to deploy adaptive AI systems to frequently collect large amounts of data and quickly adapt to changes and differences.

"But it is not easy to achieve fully automated decision-making, and the factors affecting enterprise decision-making in the future will become more and more complex, and these factors will interfere with the decision-making intelligence model and affect the correctness of the final decision, so a more flexible and powerful augmented AI system is needed to deal with these complex factors to help automate the realization of decision-making intelligence." Ingo Mierswa, senior vice president of product development at Altair RapidMiner, said.

Reporters learned that in order to solve the bottlenecks and challenges of AI+ data in scenarios such as finance, automotive design and production, and industrial manufacturing intelligence, many technology platforms have looked for technical solutions.

"From a global perspective, there are a large number and variety of data service providers that empower the digital transformation of enterprises, including not only service providers that started with digital transformation in specific industries such as industry, finance, and retail, but also general-purpose AI technology and data analysis product providers. Under the background that data-driven business development and intelligent decision-making have become an important trend of enterprise digital transformation, how to efficiently use the massive and complex data accumulated by enterprises, mine and give full play to the greater value of data, and open up the data circulation and digital transformation of the whole process of the whole life cycle such as enterprise design and development, manufacturing, marketing, sales, and operation and maintenance, has become a new business challenge that enterprises urgently need to answer at this stage. Liu Yuan, general manager of Altair Greater China, told reporters. Although many enterprises in various industries are striving to achieve full-process data-driven, there are still silos between departments and personnel, making it difficult for enterprises to correctly and efficiently use rapidly growing data, resulting in a variety of "frictions" in the application of AI technology and AI products. The "friction" in these data analysis will become an unstable factor in the digital transformation process of enterprises, resulting in project failure, cost and waste of personnel investment.

Xiao Yanghua, director of Shanghai Key Laboratory of Data Science, pointed out that data intelligence such as data + AI is an intelligent form with data as the analysis content and knowledge discovery and application as the core content, in order to enable machines to have high-order human cognitive capabilities, including understanding, reasoning, decision-making, interpretation, planning, etc. In this case, the continuous interaction between data-driven (domain pre-trained large models) and knowledge-driven (dynamic knowledge graph) dual systems is the key to realizing data intelligence. Low-code, plug-in, collaborative, explainable, high-performance, and whole-process coverage are the functional requirements of data intelligence. Complex decision-making is the goal of data intelligence.

"With the evolution of AI technology and changes in demand, data intelligence application scenarios have gradually changed from large-scale simple application scenarios such as smart travel and intelligent search to small-scale complex application scenarios such as smart medical care and smart industry, and gradually developed from people-centered to human-machine-object equal attention. To truly play the role of data intelligence in empowering enterprises, the deep integration of data and domain knowledge will be a necessary condition. He emphasized.

For more information, please download 21 Finance APP