Whenever the AI wave emerges, the financial industry, as a typical representative of data-intensive industries, is always at the forefront of the era.

The latest case is undoubtedly the large model. In this new round of productivity revolution led by AI large models, various new product forms are advancing rapidly. The powerful content understanding and generation capabilities of large models have promoted the development of AGC (Generative AI) technology, making it possible to create more intelligent and personalized AI digital humans.

AI digital human technology has thus ushered in a major development opportunity, and it is expected to bring breakthroughs in solving pain points in marketing, customer service, risk control, compliance, and other aspects of the financial industry.

So, how can financial institutions create AI digital humans that are suitable for themselves? In this regard, leading financial institutions and technology companies have already conducted active exploration.

As an online investment decision solution provider, Jiufang Wealth (09636.HK) recently launched the digital human "Jiuge", which can to some extent play the role of a professional human investment advisor. For example, it can answer various securities investment questions through multi-round conversations, track the dynamics of the securities industry, select stocks, provide data support, and assist users in making investment decisions.

AI digital humans are deepening their involvement in the financial industry in a new way.

The surge of large models towards investment advisors

In order for AI digital human technology, including the underlying large models and AGC capabilities, to truly land, finding suitable application scenarios is the key.

Understanding the core capabilities of AI digital humans is a crucial step in "job hunting" for them. And having creative conversational abilities is a breakthrough capability that generative AI demonstrates compared to traditional AI.

Traditional AI often lacks understanding of the context when answering questions, resulting in low relevance of the answers and mechanical expression. On the other hand, generative AI can understand longer contexts and engage in human-like thinking and conversation. In terms of "creative" abilities, traditional AI can only output answers based on preset tasks (such as classification and numerical prediction), while generative AI can automatically generate coherent graphics, text compositions, and even code.

Based on this, combined with the virtual image that resembles a real person, AI digital humans have the potential to be comparable to real humans. And large models can provide them with a wider range of knowledge and information support, thereby improving their application capabilities in financial scenarios. AI digital humans can be more professional.

From the demand side, investment advisors may be the most "in-demand" industry in the financial sector during the era of wealth management transformation. According to data from the China Securities Association, as of June 10th, there were 74,570 investment advisors in the Chinese securities market, while the number of A-share investors during the same period was about 218 million. This means that one investment advisor has to serve over 2,900 A-share investors.

On the other hand, the overload of covering a large customer base makes it difficult for investment advisors to have a deep understanding of their clients, and personalized services are hard to come by. Many investors can only accept general investment portfolio advice and cannot make more accurate investment decisions based on their own needs.

It can be said that under the backdrop of the accelerated transformation of the entire wealth management industry, the number of licensed investment advisors in China is far from meeting the demands of a large number of investors.

In addition, the lack of technological support is another major shortcoming of investment advisors. Traditional advisory services mainly rely on manual consultation and analysis, which leads to low service efficiency in the advisory industry and cannot meet the demand of clients for real-time and accurate investment advice.

In this regard, intelligent robo-advisors based on technologies such as big models and AIGC have the potential to play an important role in addressing the gap in investment advisors.

However, this innovation still faces a series of pain points, including high technological barriers leading to high production and operation costs; at the technical level, it is unable to truly approach the real, natural characteristics of humans; high requirements for specialized knowledge in the financial industry, digital advisors need to continuously update the content of the corpus and undergo model training, which is quite challenging.

What can intelligent robo-advisors bring?

Under the joint action of supply and demand, intelligent robo-advisors have finally emerged.

The first one to "eat the crab" is Jiufang Wealth. This Hong Kong-listed company recently launched the digital advisor "Jiuge". This intelligent robo-advisor, developed by Jiufang Wealth, Huawei Cloud, and iFlytek, can already play the role of a professional human investment advisor to a certain extent.

According to data from the Securities Association, Jiufang Wealth, known as the "number one stock investment education company in Hong Kong", has over a thousand employees with securities industry qualifications and 251 employees with securities investment advisor qualifications. As of June 2023, the number of investment advisors in Jiufang Wealth ranked first among the 79 companies with securities investment advisor qualifications.

In the evening of August 8th, Jiufang Wealth showcased a series of capabilities of "Jiuge" at the press conference. I also personally experienced many functions of the digital assistant on the "Jiufang Diagnosis" App. In summary, Jiufang Intelligent Investment Advisor "Jiuge" has eight core capabilities: market analysis, sector exploration, hot spot tracking, stock diagnosis, strategy generation, financial encyclopedia, event reasoning, and emotional companionship.

Digital assistant "Jiuge"

For example, how to solve the problem of investors not having time to monitor the market? "Jiuge" can provide investors with real-time index trends, abnormal sectors and stocks, northbound capital trends, and other market dynamics, offering a panoramic market and industry trend analysis of A-shares.

Another example is, with the emergence of new technologies and concepts such as "room-temperature superconductivity" and "big models", why does a certain event have a huge impact? "Jiuge" can integrate knowledge content such as logical maps and industry chain maps, deeply analyze hot news and major events, and explore investment opportunities.

Also, what should investors do when they always impulsively sell their stocks? "Jiuge" can use various interactive dialogue methods to identify and understand user emotions, provide psychological support and encouragement, help users deal with market fluctuations, and send risk warnings at critical moments to avoid investment mistakes caused by emotional fluctuations. It plays the role of emotional companionship.

It can be said that each capability directly addresses the specific demands and pain points of investors.

The CEO of Jiufang, Cai Zi, said that "Jiuge" divides the market analysis system into four mainstream analysis systems: technical analysis, fundamental analysis, capital analysis, and news analysis. It covers dozens of analysis dimensions and nearly a thousand stock diagnostic factors, scanning all listed companies in the market in a 360-degree comprehensive manner. At the same time, "Jiuge" can combine mainstream analysis dimensions and Jiufang's unique indicators for in-depth analysis, and provide diagnostic conclusions to investors through natural and realistic interaction experiences and simple and easy-to-understand expressions.

In fact, Jiufang Wealth has long been a pioneer in the digital transformation of the securities investment advisory field.

In 2019, the first-generation text dialogue robot "AI Wangcai" was released; in 2020, a "package" of artificial intelligence tools was launched, including business risk insight, Jiufang OCR model, financial false public opinion recognition, user sentiment analysis, financial entity recognition and error correction models; in 2021, AI Inspector and AI Monitor were launched to achieve integrated intelligent compliance review; in 2022, the "Intelligent Creation Workshop" was self-developed, realizing one-click intelligent creation.

Exploring AI Digital Humans

Digitized humans are not a new concept in the industry. Highly realistic virtual images have already been achieved in the concept of the metaverse. However, the creativity of the digital human "Jiuge" lies in its reliance on the capabilities of Jiuzhang Securities' large-scale models in the securities field.

It can be said that audiovisual synthesis, lip synchronization, and other digital human interaction technologies create its "physical body", while the large-scale models in the securities field give it a "soul", which is even more crucial.

However, there are two major problems standing in front of the financial large-scale models.

On the one hand, compared with other industries, the securities industry has higher requirements for the stability, accuracy, data security, and compliance of large-scale models. The large-scale model solutions represented by ChatGPT obviously cannot achieve "error-free".

On the other hand, it is the question of whether to use large-scale models or small-scale models, or in other words, whether to use general large-scale models or industry-specific large-scale models.

For existing general large-scale models, the question is how to connect and make them work, which is a question of "usage". However, if the large-scale model itself has limitations and cannot meet the needs of the securities industry, and these limitations cannot be overcome through "usage", then it is necessary to build or promote the construction of a large-scale model specifically for the securities industry.

In the case of the digital human "Jiuge", its core is the large-scale model in the securities field of Jiuzhang Securities. This large-scale model for the securities industry is derived from the iFLYTEK Starfire Cognitive Large-scale Model, which combines industry-wide knowledge bases such as macro data, market data, information data, announcement data, as well as in-depth research reports, special indicators, policy interpretations, market analysis, hot topics, course teaching, chief stock diagnosis, announcement mining, and other Jiufang characteristic knowledge bases. It has undergone large-scale model training to enhance securities industry knowledge.

In short, a more professional securities industry large-scale model has been forged based on a general large-scale model.

And Huawei Cloud plays a particularly crucial role in this step. As we all know, the cost and difficulty of building and training large models are high, and the flexible and elastic Huawei Cloud intelligent computing platform provides computing power guarantee for the training of these large models. The pain points mentioned earlier in the innovation of intelligent investment advisory digital human, such as high technical barriers and difficulty in model training, are also solved.

In this sense, the core value of the digital human "Jiuge" lies in not only demonstrating the huge potential of digital humans in the field of securities large models as a successful application case. The cooperation between Jiufang Wealth, iFlytek, and Huawei Cloud also demonstrates the feasible path financial institutions should take when exploring large models and applying digital humans.

Ultimately, the mission of digital humans is to return to the essence of the financial industry, effectively solve users' financial needs, and deliver information and provide services to users in a more intelligent, secure, and accurate manner. AI digital humans empowered by large models represented by "Jiuge" are striving to approach this essential demand, and they will undoubtedly be more widely applied and develop in the investment advisory industry, and even reshape the industry ecology. (This article was first published on the TiPost App, author | Cai Pengcheng)