At the 2023 World Artificial Intelligence Conference (WAIC) held from July 6 to 8, the reporter noted that with the continuous enrichment of digital technology coverage scenarios, generative AI and large model applications have gradually landed in many fields. While new technologies continue to reshape thousands of industries and help their digital and intelligent transformation, they are also accompanied by a series of new thinking and new challenges.
"The training and application of large models require massive data and high-quality datasets, and the current datasets are not only limited in scale and inconsistent in quality, but also have huge computing and storage needs for large models, and the demand for hardware equipment and technology is also very high." In addition, protecting user privacy and data security has also become an important issue. Jia Hao, vice chairman and secretary general of Zhongguancun Digital Intelligence Artificial Intelligence Industry Alliance, said at the meeting that digital intelligence transformation not only needs to respond to the current market demand, innovate in business processes, organizational structure, system platform, etc., but also need forward-looking layout to seize new market needs.
Multi-scenario applications
In recent years, artificial intelligence technology has continued to upgrade and iterate, with technological breakthroughs such as intelligent speech and natural language processing, generative AI and large models have poured in, achieving close integration with all walks of life. Among them, intelligent customer service has undoubtedly become one of the most potential scenario applications.
According to the "2023 China Intelligent Customer Service Market Report" released by Sullivan, the scale of China's intelligent customer service market will reach 6.7 billion yuan in 2022, and with the continuous expansion of the boundary of intelligent customer service application scenarios, and the continuous expansion to marketing, sales and other scenarios, it is expected that the market size is expected to grow to 18.1 billion yuan by 2027, and the market is showing explosive growth.
Sullivan and Yuan Xucong, chief analyst of Toubao Research Institute, said at the meeting that especially the emergence of AI models this year will help break through the bottleneck of "intelligent customer service is not intelligent enough", achieve more intelligent and efficient and personalized customer service products, and help enterprises reduce costs and increase efficiency. For example, in terms of conversation analysis insights, the AI model can mark communication records, actively analyze agent work behavior, and automatically generate best practice SOPs (support-oriented processes), greatly tapping the potential value of a large amount of conversation information in the call center.
In Yuan Xucong's view, by combining data availability, technology maturity and application degree, intelligent customer service currently has a high degree of application in the fields of finance, e-commerce, consumption, and government affairs, of which more than 90% penetration has been achieved in customer service business scenarios in the financial and e-commerce fields. The realization of this high penetration is mainly due to the development of AI technology and the optimization of technology in the customer service business. Taking financial scenarios as an example, intelligent customer service systems have become a standard product of bank customer service centers, and have been widely used in customer service and marketing, risk control management, process optimization and customer experience improvement.
At the same time, the rapid development of new technologies has also brought more possibilities for the development of office intelligence. During the exhibition, language intelligence technology company Midu released the first software and hardware integrated localization knowledge Q&A and content generation big language model - "honeynest knowledge Q&A and content generation big language model".
It is reported that the model can enhance knowledge based on the documents entered by users in real time, and carry out customized knowledge Q&A and content generation for relevant knowledge in the documents, so as to realize the content generation of "thousands of texts and thousands of faces", and help enterprises and individuals create exclusive knowledge Q&A and content generation models.
As the core element of enterprise operation, the financial system has also been deepening the construction of digital and intelligent capabilities in recent years. At present, many enterprises have applied artificial intelligence technologies such as character recognition, natural language processing, and machine learning to financial work such as financial accounting, management accounting, and internal control to accelerate the digital transformation of enterprises.
Zhou Ye, chairman and CEO of Huifu Payment and adjunct professor of Shanghai National Accounting Institute, used a series of business data to explain the intelligent financial application empowerment business: more than 10 billion settlement volume every day is fully automated; Nearly one million merchant invoicing is basically automated; It can query, summarize and manage funds in seconds; More than 30 million transactions are processed daily, and daily reports of various departments are automatically generated.
Liu Yuezhen, former chief accountant of China National Petroleum Corporation, took PetroChina's intelligent global financial sharing service platform as an example, and said, "Our original financial sharing scheme employed 7,100 people, and now it is expected that only 3,300 people are expected, and only 2,000 people are actually in place.
Specifically, its Gansu sales company shared 215 financial personnel before, and after sharing, the pressure was reduced to 83, a decrease of 60%, and 132 people were transferred to front-line operation positions and other departments; Changqing Oilfield shared 1,100 financial personnel before sharing, but only 630 after sharing, and the entire financial personnel decreased by more than 40%.
The challenges go hand in hand
While the iteration of AI technology innovation accelerates the trend change of diversified application scenarios, it also brings many new ideas and challenges to the development of all walks of life.
From scratch, intelligent financial applications, which are new things, pose new challenges to accounting policies and regulations.
"Many of our current financial application scenarios have used RPA and IPA applications, will the internal control undergo great changes, and what are the risk points?" Who will be responsible for the problems, and should the new regulations reconsider these issues? Liu Qin, dean, professor and doctoral supervisor of the Institute of Intelligent Finance of Shanghai National Accounting Institute, said at the meeting that the development of intelligent finance is a complex system engineering, which requires continuous change of the traditional management system and its development ecology, which must be a process of continuous change and continuous running-in, and must be made in many aspects such as innovation to support theory, track the latest technology, explore best practices, develop advanced products, benchmark against the world's frontier, revise policies and regulations, cultivate first-class talents and follow ethical norms.
For practitioners, in the past, a lot of work of financial personnel was placed on the processing and accounting of various vouchers and bills, as well as the completion of various statements and reports. Zhou Ye believes that after automation, financial personnel can be freed from a large number of heavy and trivial affairs, put more energy into the business, and become participants in the company's strategy.
In Zhou Ye's view, the real key to financial digitalization lies in connection, data, collaboration and ecology. Although many enterprises currently have built various systems such as business, payment, finance, treasury, and taxation, how to open up the interaction and connection between different systems, eliminate "information islands", and realize data sharing and collaboration is becoming a common problem faced by enterprises in the process of digital transformation. As a new role in the digital ecosystem, digital integrators can solve problems such as software intercommunication and data silos through new tools and methodologies, and realize system connection, data integration and process collaboration.
Zhou Ye believes that the advancement of payment is an indispensable and important element in the construction of intelligent financial ecosystem, not only with flexible and configurable payment capabilities, providing comprehensive capital and fiscal and tax management services, but also providing digital integration services such as solution customization, SaaS connection, data integration, intelligent analysis and decision-making. The construction of the digital ecosystem of smart finance requires the establishment of API open standards as soon as possible, the formulation of open rules for public domain data, the guidance and support the development of digital integrators, and the construction of intelligent financial ecological alliances.
In the field of intelligent customer service, Yuan Xicong believes that in recent years, the penetration rate of intelligent customer service in various industries has continued to increase, and intelligent customer service manufacturers have built moats through product services, data, technology and other aspects. However, market competition not only involves technical applications, but more importantly, a deep understanding of business logic, whether enterprises can precipitate industry data, broaden product boundaries, and create intelligent customer service solutions for full-link service requirements.
"By building a FrostRadar model evaluation system, we continue to track the competition in the intelligent customer service market. Insight found that leading echelon enterprises, rich experience in product delivery and operation, and industry data resources in multiple industries; At the same time, leading enterprises have a wide range of product lines and strong technological innovation capabilities, and invest a lot of resources in R&D and product innovation. Yuan Xicong said that in the fierce market competition, the core barriers of intelligent customer service manufacturers are still rooted in core technological innovation, industry data resource precipitation, product capabilities and delivery capabilities, especially the arrival of large models, how to improve product capabilities and delivery capabilities, which directly affects the market position of intelligent customer service manufacturers.