Taiwan Leads Initiative to Create Large Language AI Model for Financial Sector
In a forward-thinking move, Taiwan has initiated a project to create a large language model specifically for its financial sector. This development, as reported by Bloomberg, is aimed at supporting domestic institutions and navigating around the constraints of foreign AI platforms. These overseas technologies often fail to fully comprehend and adapt to local regulations and market practices.
Collaboration among Financial Institutions
Under the Taiwan FinTech Alliance, 16 financial institutions have come together to form a task force to expedite the project. The alliance has stated its intent to build an AI infrastructure that is “best suited to Taiwan’s financial industry”. CTBC Financial Holding will spearhead the project, with the Financial Supervisory Commission (FSC) overseeing it.
The project’s language model will be built on an open-source foundation as opposed to using Chinese platforms. The training process will utilize specialized data sets, including sector-specific information, regulatory material provided by the FSC, and the Digital Ministry’s Sovereign AI database. This comprehensive approach ensures the system reflects local language use and cultural context.
Project Timeline and Costs
The project is expected to cost between T$40m ($1.3m) and T$70m. Training is scheduled to commence in May. A prototype is anticipated to be ready by the third quarter, with the final version expected to be completed before the end of the year.
This initiative comes as major international banks, including JPMorgan Chase & Co. and DBS Group, continue to reap the benefits from their early adoption of AI. James Chen, the chairman of CTBC Bank, has noted a growing gap between “AI-ready” companies and those that are lagging. The former group is experiencing stronger revenue growth and market valuations.
Challenges and Future Plans
The Taiwan FinTech Alliance has acknowledged that general-purpose models often encounter issues in financial use. These problems include weak localisation and difficulty in matching domestic knowledge. The project’s initial phase will focus on banking, with plans to expand to insurance and securities at a later stage. By 2027, the alliance plans to introduce a FinLLM-based AI agent to assist with complex financial work and customer service needs.
Taiwan’s move to create a large language model for its financial sector signifies a significant step towards embracing AI technology at a national level. This initiative not only aids in enhancing the efficiency of financial institutions but also fosters growth and development within the country’s tech sector.
For more detailed information, click Here.
