Technology must be used to solve real problems in specific industries.
Author | George
Editor | Jing Yu
In September, on the bank of the Huangpu River in Shanghai, at the Inclusion Bund Conference held by Ant Group, Dr. Michael Jordan, an international artificial intelligence expert and namesake of the basketball superstar, explained to people in simple terms that in addition to answering people’s whims, How can the AI technology represented by large models achieve higher credibility and feasibility in solving practical problems?
Solving practical problems is the first step for cutting-edge technologies represented by AI to truly come to fruition. And this step has actually just begun.
At the Inclusion Bund Conference, Ant Group announced the self-developed “Ant Basic Model” and the “Ant Financial Model” that is deeply customized on this basis. After half a year of internal testing, two products, Zhixiaobao 2.0 and Zhixiaozhu, based on the Ant Financial large model, can provide specific help to ordinary financial participants.
"This is not a new technology, but a whole new world."
Jing Xiandong, chairman of Ant Group, quoted a line from the most popular movie "Oppenheimer" at the conference roundtable forum to describe the potential of technology to bring huge changes to business. Prior to this, technology has entered thousands of households from payment methods, payment scenarios and other consumer fields, allowing ordinary people to enjoy the value of inclusiveness; after the arrival of the " GPT revolution " , artificial intelligence technology will continue to be more professional and more The idea of security penetrates into every industry .
All this happened quietly in Shanghai in September. This city used to be world-famous for finance, but now it is accelerating to welcome the "entry" of technology.
01
Ant large model,
Solve "real problems"
Discussions on large-scale language models almost ran through the entire forum session of this year's Bund Conference. While making the entire conference full of "technological style", it also confirmed the importance that the industry and academia attach to this AI revolution and their expectations for its possible impact on production and life.
But unlike the "GPT craze" in the technology field in the first half of the year, discussions on large models by the Huangpu River in September focused more on "how to implement them in specific industries." This is an inevitable trend in future development - a large model cannot always be a dialogue or code generation machine, and specific value can be achieved only by solving real-life problems .
Specific to the financial industry, Tan Yan, managing director and global partner of the Boston Consulting Group (BCG), said in the "Generative AI Application Report in the Banking Industry ( 2023) " released in early August this year: "Generative AI is currently common in the market. The model has the characteristics of universality, cross-industry use, and fuzzy language. However, it is difficult to meet the banking industry's high requirements for financial professional capabilities and accuracy. Therefore, how to make the generative AI model "speak professional words" and "tell the truth" ’ have become two key challenges for the large-scale application of generative AI in the banking industry. In addition, the strict requirements for data security of banks and other financial institutions also mean that the fine-tuning and application of the model are likely to need to be carried out locally.”
This conversation accurately pointed out two key points for the implementation of large models in the financial industry: first, it must be more professional; second, it must be safer .
Professor Xiao Yanghua, a doctoral supervisor at Fudan University and director of the Shanghai Key Laboratory of Data Science, used the term "last mile" to describe the distance from general large models to industry large models at a sub-forum on September 8. "General large models It is the foundation of knowledge. It is broad enough, but not deep enough." As the saying goes, a person who travels a hundred miles is only half a mile, and the final step still requires time and investment to continue exploring.
It was at this forum on the afternoon of September 8 that Ant Group officially released a large financial model .
According to reports, this large industry model is based on Ant’s self-developed basic large model and is deeply customized for the financial industry. The underlying computing power cluster reaches the scale of 10,000 cards. It has been fully tested on Ant Group’s wealth and insurance platforms. At the same time, Ant also released two products based on the capabilities of large financial models: the intelligent financial assistant "Zhi Xiaobao 2.0" and the intelligent business assistant "Zhi Xiaozhu" that serves financial industry experts. Among them, "Zhi Xiaobao 2.0" has been in internal testing for nearly half a year and will be launched after completing the relevant filing work.
On September 8, Ant Group officially released a large financial model at the 2023 INCLUSION Bund Conference | Image source: Ant Group
In the words of Wang Xiaohang, vice president of Ant Group and head of the financial model, the Ant Financial model aims to solve the industry’s “real problems.” Or to put it more down-to-earth, these two products can provide more specific business support to financial participants.
Zhixiaobao 2.0 is mainly for investors and can provide high-quality market analysis, position diagnosis, asset allocation and investment education companionship and other professional services; while Zhixiaozhu can provide practitioners with investment research analysis, information extraction, professional creation, Provide in-depth intelligent services in areas such as business opportunity insights and use of financial instruments.
In addition, Ant Group has also officially opened the financial AI task evaluation set "Fin-Eval" to the outside world. Fin-Eval is composed of 28 categories in five dimensions: "cognition, generation, domain knowledge, financial logic, security compliance". Together with the financial model and two major products, it forms the "1+1" of the Ant Financial model. +2 Matrix".
02
Technology advances, safety remains unchanged
In addition to new technologies, Ant has not stopped exploring cutting-edge technologies and integrating them with the financial field.
At the main forum of INCLUSION·The Bund Conference, which opened on September 7, Ant Group Chairman and CEO Jing Xiandong borrowed the classic line "This is a new world" from Nolan's movie "Oppenheimer" to express My own understanding of the AI revolution: "The impact of this wave of large models will reconstruct many business (models)."
According to Zhang Hongjiang, a consultant at the Beijing Zhiyuan Artificial Intelligence Research Institute and a foreign academician of the U.S. National Academy of Engineering, the AI revolution brought about by ChatGPT occurred after the trough of the third cycle of artificial intelligence. “This revolution is not simply a skill or A simple tool to improve productivity, but a true platform revolution.”
Ant Group Chairman and CEO Jing Xiandong (first from right) talks about the impact of artificial intelligence technology on the financial industry | Image source: Ant Group
Ants have never been unfamiliar with changes in technology.
Since the birth of Alipay in 2004, they have been integrating technology and the financial industry, hoping to solve the pain points of the industry and even the trust issues in the operation of the entire society. From secured transactions to QR code payments to facial recognition, the transformation of different payment methods has brought subversion to the consumer Internet; and in recent years, from blockchain to large models, cutting-edge technology continues to change the future of finance and business development of the world .
At the roundtable forum on September 7, Jing Xiandong used the segment of high-speed computer rental to introduce how blockchain technology is used in traditional industries.
This is a capital-heavy industry, and its operation process requires a high degree of reliance on manufacturing, but it is also inseparable from the financing needs of the financial industry. Under traditional industry operating rules, entities in the industry need to rely on their own credit for financing, which will not only make their own financing relatively difficult, but also increase the risk of asset management and control by banks and other financial institutions. Therefore, this is a relatively high-risk industry.
The emergence of blockchain has created a new operating model: the main body of financing is no longer the company, but the actual equipment . “We put the blockchain communication modules together and put them in the headquarters. By controlling the T-Box, all the data and information in the device can be transferred online." Jing Xiandong introduced.
This brings changes in three directions :
First, everyone can see changes in each node, realizing real-time asset management;
Second, the financing and lending process has become easier. Funding parties do not need to bother to judge the creditworthiness of the entity, but instead transfer funds to specific equipment, and can see the working hours and working conditions of the equipment;
Third, it can transform the industry's operating logic from "asset-heavy" to "asset-light" and release collaboration opportunities for all parties in the industry chain.
However, in the wave of technological changes, the bottom line of safety in the financial industry can never change. A very classic question is, when payment becomes easier, will fraud become more likely to occur? This is a problem that Ant, as an industry giant, must solve. It is a corporate responsibility and the foundation for the company's continued development.
Ant’s solution is: “To realize the exchange of data value, rather than the exchange of data itself.” This requires a breakthrough at the level of privacy computing. The data must be available but invisible, computable but not identifiable. Technological breakthroughs are important, and cost is an important issue that needs to be solved commercially.
Jing Xiandong said at the roundtable forum on the 7th that Ant is investing heavily and hopes that one day in the future the cost and efficiency of encrypted computing will be the same as today's plaintext computing.
"That era will truly usher in the explosion of the data age."
*Source of header image: Ant Group
This article is an original article by Geek Park. For reprinting, please contact Geek Jun on WeChat geekparkGO
Geek asked
How large models can be used to solve practical problems
Achieve higher credibility and feasibility?
Hot videos
At 10:08 am on September 8, Huawei Mate X5, a new foldable screen phone, quietly started pre-sale on Huawei Mall again. Come and take a look at the actual shots of the new phone!
Like and follow the Geek Park video account ,
Watch more great videos
