In the broader landscape of technology, a significant transformation is unfolding, propelled by advancements in artificial intelligence (AI), particularly through the innovations of large language models, multi-modal capabilities, and digital twinsThe realm of AI-generated content (AIGC) is entering a new phase of exponential growth, where the creation of digital humans—avatars capable of human-like interactions—has emerged as a critical avenue for content generationThis transition is driven by the significant reduction in production barriers due to enhanced large model technologies.

However, the adoption of digital human technologies is not without its challengesStakeholders are currently grappling with high investment costs, significant bottlenecks concerning content creation efficiency, and the complexities of client management and operations

As technological refinements in visual, auditory, and natural language processing accelerate, digital humans are poised to enhance online-offline connectivity, facilitating diverse human-machine collaboration models through inquiries, companionship, and custodial services across an expanding array of sectors.

Industry experts argue that leveraging natural language understanding and intelligent dialogue technologies to rapidly construct knowledge databases can significantly elevate the cognitive capabilities of digital humans, rendering them increasingly intelligent in delivering specialized servicesFurthermore, the advancements in digital human-related textual technologies can streamline content deployment and distribution, ultimately driving down production costs for enterprises.

The impact of large models is anticipated to redefine interaction modalities, unveiling a significant market potential for digital humans, projected to be worth billions.

The emergence of large language models has significantly enhanced the intelligence of AI technologies, moving beyond traditional applications like recognition and 2D to 3D reconstructions

• US November CPI Seals the Deal on December Rate Cut
• ETFs Leading the Tech Charge
• Nikkei 225 ETFs See Wild Swings
• Leading the Way in New Energy Vehicle Production
• Evergrande's Losses Exceed 800 Billion: Debt Restructuring Crucial

AIGC now not only has the ability to autonomously generate text and images but also to create multimedia content including audio and videoAs multi-modal AI technologies continue to evolve within the domain of digital humans, their interactive abilities are becoming increasingly sophisticated, allowing for more human-like decision-making and verbal output.

According to a report from IDC, the integration of natural language processing with multi-modal technologies enables the recognition of user emotions and needsThis results in emotional comprehension and feedback capabilities, granting digital humans the ability to automatically generate language, expressions, and actions that facilitate natural interactions with usersCurrently, amidst a surge in content generation technologies, digital humans are transitioning from an algorithm-driven phase (Level 3) to one of smart language interaction (Level 4), where human oversight is required only in specific instances.

Notably, the continuous technological iteration could pave the way for a multi-billion-dollar market for digital humans

Major tech players, including Microsoft, Baidu, JD.com, Tencent, and Douyin, are increasingly entering the digital human space, offering a variety of types, such as broadcast-style, interactive, and intelligent voice assistant modelsIt is projected that by 2023, the market for virtual digital humans in China could reach an impressive 270 billion yuan.

Despite this momentum, significant challenges remain in the implementation stages even as large models accelerate the intelligence revolution for digital humansAdvanced AI systems demand vast quantities of high-quality data for training; however, there exists a relative scarcity of large-scale, ready-made Chinese datasets, coupled with significant interoperability barriers within China's tech ecosystemAdditionally, critical power computing resources necessary for the training and operation of large models are still lacking- high-quality computational devices and infrastructure need to be more readily available.

There is a general consensus in the market that vertical industries collaborating with platforms that provide foundational computing layers, frameworks, and applications may represent the future pathway for the advancement of digital humans

Currently, with the support of cloud service providers like Baidu Smart Cloud, digital human innovations, such as Citic’s “Xiao Xin,” are already being proliferated throughout the financial service sector.

Ding Lei, the Director of the Platform Management Department at Citic’s Wealth Management Division, elaborated that previously, the training of Citic's digital human heavily relied on manual generation of materials, Q&A dialogue, and the generalization of responses to various inquiriesThe advent of large models enables the direct conversion of foundational materials, such as promotional texts and scholarly works, into the digital human's knowledge base, drastically reducing the workload associated with knowledge base construction.

In the view of Yang Wen, a Senior Analyst at IDC, the emergence of autonomous content creation will facilitate a leap in AI technologies, transitioning from understanding and integrating knowledge to actual knowledge generation, thereby improving issues such as the cost of digital humans and the narrow application scope

During this evolution, technology providers must enhance the understanding of artificial intelligence among industry users, enabling the Chinese market to swiftly embrace emerging technologies, assimilating them into practical business applications while harnessing platform tools to augment the customization capabilities of digital humans.

Breaking down training barriers facilitates accelerated implementation in financial contexts.

Despite the emerging opportunities for AI, and its pivotal role in driving enterprise transformation and upgrade efforts, the pathway to make such technology operational within industries has remained a pressing challenge that many sectors are grappling with.

Many industry insiders contend that the financial sector faces numerous challenges in its business development processes, including high operational and marketing costs for client management, inconsistent service quality, inadequate product marketing content that fails to capture customer attention, protracted content production cycles, and unsatisfactory user experiences from online robotic customer services

Digital humans, armed with knowledge database-driven natural language understanding and intelligent dialogue technology, are positioned to individually address these persistent challenges.

However, due to the high precision requirements for 3D realistic human models, the overall customization cycles are extensive, costs remain high, and the digital human corpus demands continuous updating, making their successful deployment challenging if solely relying on financial enterprises.

According to the IDC report, the synergistic fusion of digital humans and AI will forge a new model for digital transformation in the financial industryPresently, numerous financial institutions, including Citic and Bank of Communications, are collaborating with Baidu to intensify the development of digital human technologies

Notably, Citic's digital human, Xiao Xin, has been recognized in IDC’s report as a model case of best practice and exploration in AI digital humans.

Ding Lei detailed that Citic's digital human, Xiao Xin, was initially positioned as a digital wealth advisor, covering financial knowledge across banking, securities, and insurance sectorsXiao Xin is designed to support financial knowledge inquiries and facilitate wealth planning communicationsSince the beginning of this year, the application scenarios for Xiao Xin have expanded significantly, now utilized across the entire Citic financial ecosystem, including the Citic Bank and Citic Securities apps along with various online channels to deliver customer service functionalities.

It is noteworthy that Xiao Xin, in achieving human-like interactions, is equipped to deliver financial services to users in a smart, safe, and accurate manner, a feat that owes much to Baidu’s intelligent cloud technology.

On one hand, Baidu’s years of technological investment in the AI space has resulted in a digitally mature persona that responds more intelligently at the perceptual level, whereby recording merely 200 audio clips can generate an equivalent data pool of thousands of phrases

On the other hand, the digital human from Baidu has already been effectively implemented across multiple banking scenarios, such as within the Bank of Communications and SPDB, assisting human staff in conducting transactions while simultaneously utilizing industry-specific knowledge to lower service costs for enterprises.

As large model technologies continue to develop, digital humans are expected to accelerate their implementation across further financial scenariosUnlike previous iterations where digital humans depended on scripted templates to respond, Xiao Xin now possesses a comprehensive understanding of financial knowledge across banking, securities, insurance, and trust sectors, with the ability to learn and evolve continuouslyFurthermore, the "emotions" and actions of digital humans are not merely pre-configured but instead are generated through analysis powered by large models.

According to Ding Lei, whether in the context of AI advancement or the metaverse's web 3.0, digital humans are expected to serve as crucial communication conduits between individuals