The rise of DeepSeek in the financial arena signifies a transformative shift, echoing the frenzy that surrounded the initial launch of ChatGPTBanks, investment firms, insurance companies, and securities agencies are scrambling to integrate the DeepSeek series models into their operations, fueling what many are labeling an "AI arms race." This integration of Artificial Intelligence into finance doesn't just promise enhanced efficiency but also ignites a debate over job security for financial workers, often termed “financial laborers.”

As the conversation evolves, financial professionals are contemplating whether innovations like DeepSeek will make certain roles obsoleteHowever, DeepSeek posits a different narrative: rather than replacing human employees, AI technologies will catalyze an evolution towards “AI-enhanced professionals.” Those who can thrive on the synergy between human intelligence and machine capabilities, equipped with a comprehensive understanding of both business acumen and technological proficiencies, stand to outpace their counterparts in career growth.

“DeepSeek represents empowerment, not disruption,” stated one finance industry participant in an online forumThis sentiment resonates strongly, particularly in light of the financial sector's stringent compliance and security standardsThe prevailing skepticism stems from the inherent unpredictability of large-language models, necessitating human oversight in the verification of AI-generated outputs, especially when engaging directly with clients.

The race for AI dominance in financial services is intensifying.

Recently, DeepSeek unveiled its V3 and R1 models, characterized by their low-cost implementation, high performance, and extensive adaptabilitySuch features have substantially lowered the barriers for companies aiming to deploy large models internallyMajor tech platforms, including renowned names like Huawei Cloud, Tencent Cloud, Alibaba Cloud, Baidu Cloud, and JD Cloud, were among the first to integrate DeepSeek, aiming to leverage its capabilities for their diverse business strategies.

As the wave of integration gains momentum, a variety of companies and institutions are beginning to utilize DeepSeek in real-world applications

Advertisements

The financial sector, known for its robust embrace of technological advancements, is at the forefront.

A recent post from Hai'an Rural Commercial Bank on its official WeChat account declared, “DeepSeek, you understand Hai'an Rural Commercial Bank too well." This declaration underlines DeepSeek's analytical prowess, dissecting aspects such as capital strength, market share, service quality, risk management, product diversity, corporate responsibility, technical support, and employee competence—essentially performing a self-promotional review.

Whereas Hai'an Rural Bank is leading a conversation around its capabilities, Jiangsu Bank is taking a more direct approach to inform the publicThe bank announced its successful local deployment and fine-tuning of the DeepSeek-VL2 multimodal model and the lightweight R1 inference model via its “Smart Xiao Su” language service platformThese models are now integral to their smart contract quality inspection and automated valuation reconciliation processesSince adopting the DeepSeek model, “Smart Xiao Su” has significantly improved its capabilities in handling complex scenarios, showcasing enhanced efficiency in processing tasks.

Industry insiders have praised the R1 inference model for its robust performance, enabling it to tackle intricate financial data and multi-tasking environments effectivelyIts lower training costs and higher cost-performance ratio make it an appealing choice for applications in intelligent customer service, robo-advisory, and risk management.

According to data shared by Jiangsu Bank, the implementation of the R1 model, coupled with enhanced email gateway processing, has enabled fully automated workflows, achieving a remarkable success rate of over 90% in tasks such as email classification, product matching, trade entry, and reconciliationBy streamlining operations, this innovation potentially reduces manual workload by almost ten hours daily, thus amplifying operational efficiency.

Moreover, representatives from major state-owned banks have expressed keen interest in incorporating DeepSeek's open-source capabilities to create tailored applications across various domains, such as intelligent investment advising, customer service enhancement, risk monitoring, and compliance management.

The enthusiasm extends into public mutual funds, as over ten major companies—including Huatai-PB, Fortune Fund, and Nuon Financial—have announced their own applications of DeepSeek models recently.

Specifically, Huatai-PB affirmed its successful deployment of the DeepSeek series for core activities like investment research, sales strategies, risk compliance, and customer service

Advertisements

Similarly, Nuon Fund launched its own “Nuon AI Assistant,” built upon mainstream AI frameworks, targeting prime use cases in research analytics and customer relations.

In the insurance sector, Ping An has maintained a long-standing commitment to AI and big data advancements, aiming for a comprehensive digital transformationThe company is actively pursuing research and deployment of integrated big data platforms, focusing on enhancing synergies within their financial services and elder care business ecosystems.

Moreover, major brokerage firms like Guotai Junan, Guojin Securities, GF Securities, Industrial Securities, and Huafu Securities have formally reported successful projects involving DeepSeek R1's local deploymentGuojin Securities, for instance, is targeting diverse applications for information retrieval, document management, sector research, and market analysis, with plans to further leverage the technology for smart services and investment analytics.

As financial technology firms pivot towards deeper integration of AI, last weekend Financial One Account launched its own intelligent platform, incorporating models from DeepSeek and Tong Yi Qian Wen, offering comprehensive AI solutions tailored to the banking sector.

Analysts are predicting that the deployment of local models within financial firms will become commonplaceSuch strategies could enable banks to combine locally stored data with large models, thus developing proprietary models that cater specifically to their operational scenarios.

According to analyst Li Bolun from Guotai Junan, the specificity of the financial domain often necessitates bolstered data security measuresThe newly released DeepSeek-R1 allows financial institutions to operate top-tier models at a fraction of the cost while crafting bespoke models that uniquely address their needs.

For IT companies in finance, opportunities abound as they help financial entities streamline their massive data reservoirs, converting this information into structured formats for model tuning and specialized implementations.

AI large models are increasingly prioritized as a foundational component of the new digital infrastructure in finance, particularly in banking, where heightened operational efficiency, nuanced risk management, and superior customer experiences are required

Advertisements

Conventional technological solutions have proven insufficient, yet the emergence of AI large models presents fresh avenues for progressShih Wenbin, Director of AI Products at Financial One Account's Intelligent Voice Team, asserts that the V3 and R1 models are built on a multi-expert architecture, consistently performing well across various metrics.

For example, the R1 model excels in high-performance inference and extensive learning capacity, particularly regarding mathematical and technical domainsThese capabilities are particularly useful in banking scenarios that demand intensive data processing and nuanced decision-making.

In the realm of risk assessment, the R1 model adeptly synthesizes multifaceted customer data for precise evaluation, providing thorough insights into credit and market risksIts application would also afford smarter customer service through enriched understanding of client intentions and a greater ability to unearth latent customer needs, thereby recommending appropriate financial products.

Nonetheless, human expertise remains indispensable.

Despite the advances presented by DeepSeek, users have occasionally reported concerns over reliabilityInstances have emerged where DeepSeek generates content that diverges significantly from accuracy, creating potential pitfallsFor instance, the generation of fictitious academic references raises the stakes: a misinformed market analysis driven by erroneous AI data could lead to detrimental financial decisions for clients.

Furthermore, the use of large models entails substantial processing of personal and enterprise data, subsequently risking leakageIndustry insiders have raised alarms regarding the potential for misuse of that information for unauthorized activities, jeopardizing institutional reputations.

In response, DeepSeek acknowledges the possibility of heightened efficiency paired with multifaceted risks, such as data security and privacy violations, model distortions, compliance challenges, and systemic risks

Advertisements

Advertisements