As the core engine of the global economy, the financial industry is constantly changing and innovating. Under the trend of computing power, which is becoming a new productivity in the era of digital economy, the industry has launched a new round of exploration around the digitization of the financial industry.
Recently, the 2023 China International Financial Exhibition (referred to as: Financial Exhibition) was officially held in Beijing. As an international financial exhibition platform with high popularity and influence in China and even the Asia-Pacific region, the 29-year development history has also made the Financial Exhibition from the initial electronic It has developed into an annual grand event for the financial industry and related industries to showcase technological progress and product innovation, and is highly valued by various financial technology companies and major financial institutions.
As a leading provider of computing power infrastructure and services, Super Fusion was invited to participate in the exhibition with the theme of "Intelligent Finance, Computing the Future". During the conference, it released the latest series of ISV solutions with four partners. Among them, Super Fusion joined hands with Impulse Online, a leader in privacy computing and trusted artificial intelligence, to launch a trusted AIGC computing joint solution. In the rapid tide of AI development, it has thrown a "trump card" for the digital transformation of the financial industry. .
AIGC and the "second half" of digital transformation in the financial industry
In recent years, with the continuous improvement of digital infrastructure construction and the accelerated implementation of commercial applications, the development of the artificial intelligence industry has entered the "fast lane". As the AIGC boom represented by ChatGPT continues to sweep the world, the financial industry is also constantly thinking about how to accelerate the transformation and development of its own business through the capabilities of AI.
In fact, as a basic industry of the national economy, the financial industry has always been at the forefront of industrial digitization and intelligence. Because AI has the advantages of high efficiency, low cost, low latency, and easy iteration, the financial industry has launched many explorations in financial scenarios such as intelligent customer service, intelligent investment advisory, anti-fraud, and automatic business batch processing, but the results have always been unsatisfactory.
Only in terms of smart customer service and smart investment advisory, many institutions have increased investment in AI customer service on the one hand, and streamlined the team of human customer service on the other hand. As far as the actual landing experience of users is concerned, AI customer service can only give standardized answers to simple questions but solve them. No problem, the interactive logic full of slots has made countless users "want to cry but have no tears".
With the introduction of AIGC capabilities, the financial industry is ushering in new changes. In the field of intelligent customer service, Gartner believes that 85% of the interactions between customers and enterprises do not require human participation. At present, leading banks are at the forefront, and insurance and securities are also in the process of upgrading AI product replacement and assistance; in the field of intelligent investment, generative AI can be used as Auxiliary information is integrated into the recommendation system, data analysis and research reports are generated, and personalized investment suggestions are proposed, which can improve information utilization and enhance recommendation effects. With the rapid iterative development of GPT-like pre-training models, generative AI technology is expected to accelerate in the financial industry application.
In general, the goal of digitalization in the financial industry is to improve the efficiency of financial services, reduce operating costs, and provide users with more personalized and convenient financial services through the use of digital technology. The C-end user group has accumulated massive data and rich application scenarios. It is a high-quality AIGC landing scenario. However, for the high-value scenario of the financial industry, if the evolution of AIGC is to promote the evolution of intelligent services, it is not only necessary to consider the infrastructure of computing power The changes in the data security and user privacy industry pose new challenges.
Therefore, in the second half of the digital transformation of the financial industry, facing various pain points and challenges, the trusted AIGC computing joint solution jointly launched by Superfusion and Impulse Online came into being.
A Leap of Efficient Synergy of Software and Hardware
The development of AIGC is inseparable from the basic framework of computing power, algorithms, and applications. Based on the continuous deep cultivation of AIGC's full-process technology, the trusted AIGC computing joint solution jointly launched by super fusion and Impulse Online is the first to use privacy computing technology. The end-to-end privacy and security between multi-party data, computing power, and models provides "out-of-the-box" solutions for the financial industry, greatly reducing the difficulty for users to build AI platforms and the technical threshold for development, and accelerating AI innovation.
From the perspective of business scenarios, when using the AIGC model, adding privacy computing can not only use external data to fine-tune and optimize according to its own needs, but also make the AIGC model better serve the specific business and needs of the enterprise, and improve the accuracy and efficiency of the model. As a result, for external data, the scale and diversity of the enterprise's internal data sets can be expanded, thereby improving the generalization ability of the model.
From the perspective of software product architecture, the momentum trusted AIGC platform consists of six parts, of which trusted data nodes can realize external data access platforms, and combine data watermarking, data blood relationship and other technologies to confirm the data usage process Rights tracking, providing data raw materials for the data crowdfunding stage. Trusted computing power nodes provide a decentralized computing power sharing market, manage mainstream CPU and GPU computing power resources at home and abroad, and have intelligent computing power scheduling strategies to provide optimal resource support solutions for computing power scheduling. The AIGC model at the upper level is encapsulated by the platform into a unified standard reasoning instance, and the platform schedules reasoning tasks, and realizes cross-platform deployment, monitoring and tuning through cloud native technology, which reduces the difficulty of installing and deploying reasoning services and enhances model training For task stability, the AIGC service hosting module not only provides basic AIGC work management capabilities, but also provides model parallel optimization, including data parallelism and pipeline parallelism.
In terms of hardware server building, the super fusion basic platform builds an industry-leading trusted server based on a trusted execution environment, which not only supports 8 double-width GPU cards, 24 3.5-inch hard drives, but also has a one-click switch between cascaded topology and balanced topology , flexibly applicable to a variety of scenarios, FusionDirector server lifecycle intelligent operation and maintenance and other capabilities, also greatly reduce subsequent operation and maintenance costs Under the dual ecological advantages of the East and the West, software and hardware synergy are realized to ensure safer and more reliable financial delivery.
Also because of its solid foundation, the AIGC platform can guarantee the end-to-end privacy of privacy computing task data and the security of the data life cycle in credit risk assessment, financial anti-fraud, anti-money laundering, credit investigation, insurance pricing and other scenarios With the traceability of the calculation process, a complete credible computing power + AIGC solution has been formed in AIGC scenarios such as intelligent question-and-answer robots, intelligent customer service, marketing graphic generation, video generation/digital virtual humans, etc., to meet the challenges faced in digital financial practice A variety of scenarios, continuous empowerment, and provide the industry with a full-link trusted privacy computing solution.
At the press conference, Wang Pengda, product director of Impulse Online, gave an example of the application scenario of the trusted AIGC all-in-one machine in the financial industry:
Marketing graphic generation:
Taking wealth management products as an example, the traditional marketing copy generation is generally to first write the content of the marketing copy of the wealth management product, then the designer designs the picture according to the content of the copy, and the artist typesets according to the content of the marketing copy and the picture, and finally generates a one-time delivery marketing copy graphic. In the AIGC model, users only need to provide the essential information of wealth management products, select the style of the picture and the template of the distribution channel, that is, the web end or the APP end, and then start the training of the marketing model, and then adjust the model as needed Parameters, continuous optimization, and the last one-click generation of a marketing plan.
AI model open services of financial institutions:
The trusted AIGC platform assists financial institutions to open their own AI capabilities, such as OCR, NLP recognition and other services, to downstream small and medium financial institutions. Small and medium financial institutions can also upload their own customer data and complete the model on the trusted AIGC platform. Training and inference services. For example, the bank opens the contract identification model to the insurance institution, and the insurance institution can connect the customer contract data to the platform and complete the contract comparison service on the platform. Completion of computing power can ensure that customer data will not be placed on the platform side, and the content of the bank's AI model will not be leaked.
In addition, the open source AI model built into the trusted AIGC all-in-one machine can also provide services such as intelligent customer service and virtual human construction for the financial industry. The diverse needs of the model.
In the future, during the large-scale commercialization of AIGC, Impulse Online will work with superfusion to become a "beacon of computing power" in the financial industry and help the financial industry "ride the wind and waves" in the digital transformation.