2019 Industry AI Sketch: Finance Edition
-
This article reviews the applications of AI in the financial industry and its impact on the sector.
2019 was a turbulent year for China's financial sector—
From the issuance of the "No. 175 Document" by a special task force in January, marking the official start of the cleanup of internet finance, to Huang Qifan's announcement in October that China's central bank might pioneer the launch of a digital currency, and the strict scrutiny of financial app data privacy in December.It can be said that the theme of China's financial sector in 2019 was "cleaning up fake technology and doubling down on real technology." Micro-lending companies masquerading as FinTech collapsed one after another, while advanced technologies like AI, big data, and blockchain continued to enter the commercial implementation phase.
Especially for AI, the financial industry has always been a top performer in technology adoption—
With its highly digitalized and information-based infrastructure, the financial sector finds it easier to integrate technological solutions compared to other industries.Particularly in front-office operations like customer service and product sales, which are labor-intensive, the cost-saving and efficiency-boosting effects of AI are even more pronounced.
Additionally, the financial industry is more proactive in technological experimentation, with many institutions having their own R&D departments, giving the sector a solid foundation for AI adoption without the need for extensive market education.
As a result, the financial industry naturally became a key battleground for AI industrialization in 2019.
Overall, the implementation of AI in the financial sector in 2019 progressed steadily on a solid foundation. Many people could intuitively feel the convenience brought by technological advancements, even as end-users. In front-office financial operations, "verification" is almost a mandatory step, involving the confirmation of personal, invoice, and document information. These tasks, often related to proving one's identity, consume significant labor resources.
From a technical perspective, solving these problems isn't complex. Technologies like facial liveness detection, OCR recognition, and image recognition models can automate verification processes, enabling remote transactions with just a photo upload via a mobile device.
However, the challenge of applying AI for verification in finance isn't just about technical capability:
First, the financial industry inherently demands higher standards for data security and privacy. For instance, the Ministry of Industry and Information Technology's crackdown on app data permissions at the end of the year led to the removal of many banking and financial apps, primarily due to data security concerns.
Second, as mentioned earlier, the financial sector's digital and information infrastructure is more advanced than other industries. The China Banking Regulatory Commission's "13th Five-Year Plan for Financial IT Development" stipulated that by the end of the plan, all critical internet-facing banking systems must migrate to cloud platforms, with at least 60% of other systems following suit. This triggered a widespread cloud migration trend in the financial sector from 2016 to 2017.
With cloud infrastructure in place and high data security requirements, the challenge for AI adoption shifted from technical capability to deployment methods. While other industries might simply integrate APIs for facial or text recognition, such approaches are often too rudimentary for finance.
Thus, two major trends emerged in 2019:
One was tech providers adjusting their cloud solutions to meet the financial sector's unique needs through private or hybrid cloud deployments, ensuring seamless integration with existing digital frameworks.
The other was financial institutions opting to develop or procure technology to upgrade their cloud platforms with AI capabilities. Whether due to the financial sector's growing efficiency in adopting AI or tech companies' improving service offerings, these trends significantly accelerated AI adoption in finance.
As a result, most banks and fintech products now use facial and image recognition for identity verification, enabling remote account openings and reducing risks like identity fraud.
The breakthrough in AI deployment in 2019 was pivotal because it opened the floodgates for a continuous influx of technologies beyond just recognition algorithms. This led to broader applications, such as:
Smart Customer Service
Customer service, a labor-intensive function across industries, is a natural target for AI. In finance, smart customer service isn't limited to sales or post-sale support but is also widely used in debt collection.
If measured by the ratio of AI agents to human employees, collection agencies likely have the highest "AI density" in finance.
Debt collection relies heavily on phone calls, where AI agents interact with users, analyze speech via big data, categorize users, and assist human agents in decision-making. This not only boosts efficiency but also enhances workflow stability and control, revolutionizing collection practices.
RegTech (Regulatory Technology)
In a year dominated by regulatory scrutiny, AI-assisted financial regulation gained traction under the term "RegTech."
RegTech combines various AI technologies. For example, Australia's ASIC and Singapore's Monetary Authority use big data to detect anomalous trading patterns. The Shanghai Stock Exchange employs machine learning to model investor behavior and identify violations. The Tokyo Stock Exchange uses Hitachi's AI to spot market manipulation.
RegTech enables AI to not only improve individual firms' efficiency but also mitigate systemic risks, addressing the long-standing issue of regulatory lag.
Innovative Scenarios
Interestingly, many tech firms in 2019 expanded beyond digital realms into physical spaces. Companies like Tencent and JD.com introduced concepts like "unmanned financial pods," deploying facial recognition and voice interaction via microphones and smart cameras.
Such hardware-based solutions eliminate the need for robust algorithms to accommodate diverse mobile devices and lighting conditions, streamlining verification processes.
These "unmanned pods" bypass banking apps, bringing AI directly to offline users who may be less tech-savvy, reducing manual workloads and standardizing operations.
The above examples illustrate the density of AI adoption and innovation in finance in 2019. IDC's "China AI Implementation White Paper" noted that the financial sector is the most active in AI adoption, with higher project volumes and maturity. This top performer has not only aced the basics but also tackled bonus questions, delivering an outstanding report card.
Looking at the accomplishments of AI in the financial industry in 2019, we can see a potential development path that may also emerge in other sectors. AI's role in industries progresses step by step—from image recognition in individual frames to the complete transformation of business logic, and even contributing to higher-level regulatory and developmental issues.
Although throughout the year, we still witnessed AI causing quite a few 'jokes' in the financial sector—such as many banks' intelligent phone customer service still being clueless and unable to understand human speech, or some small loan companies disguising P2P products under the guise of AI, blockchain, and quantum computing—despite the bumps along the way, we are still moving toward hope.