(AsianFin)—The banking sector faces significant hurdles in applying large models due to stringent requirements for data compliance, security, accuracy, and reliability.

Although initially expected to lead in large model adoption, the financial industry lags behind others like legal and HR sectors. Challenges include technological barriers, personnel shortage, and high costs, compounded by data quality issues and privacy concerns.

Bank data, including transactions and customer information, contains vast unstructured data leading to storage inefficiencies. Issues like data omission and constraints affect model accuracy, while data security concerns impede data sharing. Large models offer immense processing speed enhancement but demand substantial computing power and storage.

The scarcity of qualified talents further complicates the application. Large model experts need both artificial intelligence (AI) knowledge and industry-specific expertise, highlighting the need for comprehensive training programs. Large model talents in banks not only need to have reserves of knowledge in the field of artificial intelligence but also need to have knowledge of the vertical industry they are in. Even if they have learned large model knowledge but lack understanding of the industry, business processes, and data, they cannot be called large model talents. Hu Liming, Vice President of Tencent Cloud, said that there is a huge talent gap in AI large models, and leading institutions are currently recruiting some AI-related professionals, such as algorithm PhDs, etc.

There is a massive amount of data within banks, including transaction data, customer information data, risk control data, performance data, etc. Due to the requirements for archiving data sources, bank data contains a large amount of unstructured data, such as image scanning data.

These non-standard data not only occupy a large amount of storage resources but may also produce duplicate records. However, due to the inability to standardize integration, analysis, and utilization, it greatly reduces the operational efficiency of employees and equipment. The emergence of large models can solve this problem well, but it also requires a large amount of data for training. Due to factors such as restructuring, changes, and optimization, the data accumulated by banks may have data quality issues, including data omissions, data constraints, data security, etc.

However, large models offer transformative potential, streamlining processes and improving efficiency. They enable banks to optimize staff roles, enhancing service quality and productivity. Yet, large model adoption must prioritize talent development and customer trust, ensuring sustainable growth and mitigating workforce displacement fears.

Despite the hurdles, large models promise significant advancements in banking operations, provided they address data integrity, talent shortages, and customer concerns. As banks navigate these complexities, collaborative efforts between industry stakeholders and technology providers will be essential for successful large model integration and realization of their full potential.