The popularity of ChatGPT at the beginning of this year has set off a wave of generative AI (AIGC) in the world. Enterprises such as OpenAI abroad and domestic Baidu have all made strong deployments on AIGC.
In various application scenarios, digital humans powered by AIGC have attracted the attention of the market.
In fact, digital humans are nothing new. As early as 1964, Boeing developed the first digital human with a "human" image, that is, Boing Man. The first digital images that most young people in China come into contact with should be the virtual idols Hatsune Miku and Luo Tianyi.
With the iteration of AIGC technology, the image of virtual people is becoming more and more realistic. Not long ago, some netizens used an AI model to generate an AI digital avatar of singer Stefanie Sun. From the appearance to the sound, it was impossible to tell the truth from the fake. In addition, now that you watch short videos, you can often find some digital people who are anchors, almost exactly like real people.
There is no doubt that AI digital humans have become popular. However, many companies still don't know how to create and utilize digital humans to create greater value. If you fall behind the times, you may be eliminated.
Based on this, the International Data Corporation (IDC) released the latest "PeerScape: AI Digital Human Best Practice Cases and Explorations" report (hereinafter referred to as the "Report"), pointing out the direction for many companies. The "Report" pointed out that AI digital humans have initially landed in industries such as finance, telecommunications, and media. The current typical scenarios are intelligent customer service, virtual anchors, and intelligent marketing.
In addition, IDC also selected some representative industry cases for analysis. In particular, the financial industry is considered by IDC to be a key area of AI digital human practice. Baidu has two digital human cases, both of which are regarded as typical by the "Report", one of which is the "Xiaoxin" of CITIC Financial Holdings.
Why has the financial industry become the focus of AI digital human practice? How did CITIC Financial Holdings' "Little Letter" land? What role does Baidu play in it?
Use this report from IDC, along with in-depth market observations, to find out.
Digital Human: The "New Assistant" of the Financial Industry
Just as "Internet+" gave various industries the wings to take off, now "AI+" also brings infinite imagination.
Especially in the financial industry, which deals intensively with various industries, AIGC is exerting a strong momentum.
According to estimates, the scale of China's industrial finance is close to 300 trillion yuan, and industrial digital finance is a new track for the transformation of financial institutions in the future. It is expected that the scale will exceed 400 trillion yuan by 2025.
In essence, digital finance still has obvious characteristics of the service industry, and marketing is an important engine to promote the development of the industry. However, at present, there are still some pain points to be solved in the industry, which hinder the pace of development.
In online channels, there are pain points such as product marketing content that cannot attract customers’ attention, long content production cycle, high cost, and poor user experience of online robot customer service; in offline channels, there are problems such as high manpower input costs, low service efficiency, and inconsistent service quality, making it difficult for some marketing activities to be carried out more efficiently.
It is not that these pain points have not been discovered before, but that the technology has not yet matured enough to completely solve them. With the outbreak of AIGC, digital humans have become the "right-hand assistant" to solve these pain points.
AIGC itself is a digital technology, it needs a carrier or an external output channel, and AI digital human is one of the best choices. It can connect the virtual and real worlds, form a new form of content output, carry out brand activities more efficiently and comprehensively, and create deeper marketing value.
The "Report" shows that most digital humans will be in the L1-L3 stage before 2022, and their "capabilities" are relatively simple. They can only convert 2D pictures into 3D digital humans. This may solve the pain points of one or two companies, but it cannot be implemented on a large scale to solve larger industrial-level problems.
Technological innovation is always the best driving force for technological development. Driven by intelligent technologies such as pre-trained large models and voice semantics, digital humans are about to enter the L4 stage, that is, to realize preliminary intelligent interactions and further enhance productivity and creativity.
To put it bluntly, AI digital humans have become more like "humans" and can help the financial industry create greater marketing value.
On the one hand, technologies such as end-to-end dialogue models, speech semantics, natural processing, and AIGC can enable digital humans to have unique appearances, personalities, specialties, etc. in a shorter time, instead of giving people a cold machine feeling. This will allow financial institutions to have a more authentic and three-dimensional brand image when facing customers.
On the other hand, combined with the integration of multimodal technology, digital humans can perceive the emotions and needs of customers, and then automatically generate interactive language, expressions and actions, such as smiling, sorry, etc., thereby providing services to customers more efficiently. In this way, financial institutions can further target target customers through digital humans, gain insight into customer needs, and realize intelligent and refined marketing strategies.
The report released by IDC is quite authoritative for the insight into digital people. After all, with a development history of more than 50 years, it is already a world-leading and professional provider of market research, consulting services and exhibition activities.
Through this report, it has been predicted that AI digital humans will have greater assistance in the financial industry.
The "Baidu power" behind "Xiaoxin"
Opportunities of the times and technological dividends are driving almost all industries and companies forward, but not every player can produce faster acceleration.
Only those enterprises with sufficient innovation consciousness and practical spirit can become the trend-setters running ahead.
In the "Report", several financial institutions were selected as typical cases of digital human practice, among which CITIC Financial Holdings is particularly prominent.
As a platform company of CITIC Group's comprehensive financial services sector, CITIC Financial Holdings, like other financial institutions, also faces pain points in the process of digitization. Through the digital human "Xiaoxin" released in September last year, CITIC Financial Holdings has solved these pain points and created new value.
So far, "Xiaoxin" has created many firsts.
First, it is the industry's first digital human wealth advisor.
When customers of CITIC Financial Holdings encounter questions they don't understand, "Xiaoxin" can answer questions online, explain products, and help customers understand products and services in a more humane and efficient way. More than that, it can also chat with customers about in-depth things, such as investment portfolios, investment strategies, etc., and give corresponding suggestions.
Many customers who have been served by "Xiaoxin" said that they felt "cheerful, confident, sunny, and professional". This is much better than a simple customer service experience, and it also makes the brand image of CITIC Financial Holdings more professional and full.
Secondly, "Xiaoxin" is also the beginning of the digital transformation of CITIC Financial Holdings.
In March last year, CITIC Financial Holdings was established, and "Xiaoxin" was born in the following six months. Now it has become the online image ambassador of CITIC Financial Holdings and its subsidiaries, providing customers with intelligent services through the channels of each subsidiary.
"Xiaoxin"'s brain thinking, appearance, audio-visual perception, and language ability come from the relatively mature digital human technology of Baidu Smart Cloud and the "knowledge base" of the financial industry of CITIC Financial Holdings.
Ding Lei, Director of the Platform Management Division of the Wealth Management Department of CITIC Financial Holdings, said that as early as 2016, CITIC Group cooperated with Baidu to build intelligent robots for outlets, and there is a basis for cooperation. Therefore, CITIC Financial Holdings chose to build a digital human with Baidu, and chose the customized solution provided by the Baidu Smart Cloud digital human team.
In the process of cooperation, Baidu's technical strength, intelligent customer service team and cloud service support capabilities have all played a key role. This is also the most obvious technical advantage of "Xiaoxin" compared with other digital people.
For example, relying on Baidu Smart Cloud's natural language understanding and intelligent dialogue technology, CITIC Financial Holdings quickly built a question-and-answer knowledge base, which greatly improved the cognitive ability of "Xiaoxin", allowing it to acquire professional knowledge covering banking, securities, trust, insurance and other financial fields, and be able to take on the responsibility of wealth advisors.
Ding Lei said that in the process of building digital humans, CITIC Financial Holdings cooperated with Baidu Smart Cloud to build a digital human platform, and Baidu’s cloud services also provided technical support for digital humans. The cooperation between the two parties was very smooth.
In an NBA game, in order to win, you need a "key force".
On the one hand, this "power" can organize the coordination between the players and advance tactics more flexibly and accurately, and on the other hand, it can hit the defensive pain points and throw in the winning goal. This "key force" is usually promoted by star players, such as Jordan, Kobe, etc., who are also known as the team's "Mr. Key".
Shopping malls are like stadiums. In the process of enterprise development, they actually need "key force" and "Mr. Key".
In the process of cooperating with CITIC Financial Holdings to create a digital human, Baidu Smart Cloud plays the role of "Mr. Key" and exerts "key strength".
Future-oriented differentiated "source power"
Where does this "key strength" of Baidu Smart Cloud come from?
The "differentiated competitive strategy" put forward by Michael Porter, the "father of competitive strategy", emphasizes that in the competition, products or services must be optimized by peers in order to gain more customers and consolidate competitiveness. How to achieve differentiation, in fact, the core lies in:
What people don't have, I have, what others have, I am excellent, and what others have, I am new.
In terms of "what others don't have", Baidu's long-term accumulation in the field of AI and exclusive innovative technologies make digital humans more intelligent.
For example, Baidu's intelligent technology in the perception layer makes digital humans look and sound more intelligent. Intelligent technology at the cognitive level enables digital humans to have deeper thinking capabilities.
Take the intelligent technology of the perceptual layer as an example. In the early TTS field, digital human beings were driven by text to speak. Although people can understand it, the experience is not good. Sometimes the sound and picture are not synchronized, and sometimes the mouth shapes do not match. It is not like a real human speaking.
Baidu uses fine-grained acoustic modeling and autoregressive modeling. Improved style prediction. It only needs to manually record 200 sentences of audio corpus to generate a sound library effect comparable to tens of thousands of sentences, which sounds basically the same as real people speaking.
In terms of "people have their own advantages", Baidu Smart Cloud has accumulated rich experience and mature implementation cases for digital human financial assistants. For example, "Jiaojiao", the digital person of Bank of Communications, "Xiaopu", the first digital employee of Shanghai Pudong Development Bank, and "Xiaoan" of Guotai Junan.
They can not only assist enterprise employees to handle business, but also upgrade the industry-specific knowledge base, sometimes knowing more than real employees. This can provide professional financial services to more customers, while reducing the cost of the enterprise.
Experience is sometimes the moat of a company. Even if the peers have caught up technically, there is still a gap with Baidu in terms of experience. And this takes time to settle, and it cannot be formed in a short time.
In terms of "what people have, I am new", Baidu continues to iterate on technologies such as Wenxin large-scale models, bringing new directions for digital people.
The essential change brought by the Wenxin large model to digital humans is that digital humans in the past were relatively rigid and rigid, and could only conduct questions and answers based on pre-designed templates. When the user mentions a specific keyword, the digital human will answer it in a "reciting" style, and the question that mentions "super-class" cannot be answered.
The upgraded digital human based on the Wenxin model has become more agile and intelligent. It is no longer limited to pre-set templates, but can continue to learn and evolve. For example, in the financial industry, after continuous learning, it can become proficient in financial knowledge in all fields such as banking, securities, insurance, and trust, and become more "all-rounder".
This change has practical significance. A digital human based on the Wenxin model can be fully qualified for difficult jobs such as financial planners, even better than real financial managers. Statistics show that the shortage of financial planners in China has reached 600,000, and the Wenxin model will undoubtedly make this employment gap smaller.
Many companies are already optimistic about Baidu's Wenxin model. Ding Lei said that large models have been very popular since last year, so CITIC Financial Holdings also found that "Xiaoxin", as a digital person, must have the blessing of large models in order to better interact with users. The cooperation with Baidu is also based on Baidu's Wenxin model to study how to make digital people smarter.
In general, Baidu Smart Cloud can become the "Mr. Key" in the process of building digital human beings for many enterprises, precisely because of these differentiated competitiveness.
epilogue
The wave of digitalization is coming, and it has become an irresistible trend.
In the wave, finance and many other industries are being dragged forward. Among them, the digital human has become a powerful assistant to push them forward and can solve many difficult pain points, which has been confirmed by the IDC report.
In this process, enterprises must learn to choose the correct "Mr. Key" in order to exert their "key strength". The cooperation between CITIC Financial Holdings and Baidu Smart Cloud to create "Xiaoxin" is a high-quality and valuable reference object.