KBTG announces strategic partnership with AI Singapore and Google Research to advance LLM development in Southeast Asia under Project SEALD

27 Sep 2024

KASIKORN Business-Technology Group (KBTG) has partnered with AI Singapore and Google Research to develop Large Language Models (LLMs) in Southeast Asia. Guided by KBTG's "Human First x AI First" strategy, this partnership solidifies the company's position as a regional AI leader. Under Project SEALD (Southeast Asian Languages in One Network Data), KBTG will engage in research and development as well as exploration of world-class AI technologies. This initiative aims to bolster financial technology capabilities and deliver innovative IT solutions for the benefit of customers across the region and KASIKORNBANK, in an efficient and sustainable manner.

KBTG announces strategic partnership with AI Singapore and Google Research to advance LLM development in Southeast Asia under Project SEALD

Dr. Thadpong Pongthawornkamol, Managing Director of KASIKORN Business-Technology Group (KBTG), said, "Project SEALD (Southeast Asian Languages in One Network Data) was launched by AI Singapore and Google Research in March 2024. The project aims to evaluate, train, and fine-tune large language models (LLMs) for Southeast Asian languages such as Thai, Indonesian, Tamil, Filipino, and Burmese. This initiative involves collaboration with partners from various countries with LLM expertise to expand the capabilities of LLM and the understanding of cultural contexts, while promoting the application of LLMs for social benefit region-wide."

As the objectives of Project SEALD are in alignment with KBTG's "Human First x AI First" strategy, which focuses on human-centered AI technology research and development, KBTG has partnered with AI Singapore and Google Research in Project SEALD, one of the region's largest LLM initiatives. The expertise of KBTG personnel, combined with resources developed from THaLLE by KBTG, an intelligent LLM financial model, will help foster the project's growth and development. This collaboration also provides KBTG the opportunity to learn and work closely with leading global AI partners, positioning KBTG at the forefront of driving LLM growth in the region.

For this year's project goals, KBTG will focus on co-creating high-quality Thai language datasets, which are currently underrepresented. In the future, researchers from the lab will be assigned to participate in the development and evaluation of models in subsequent projects. The aim is to leverage the knowledge gained from this collaboration to develop financial technologies for KBank and other regional businesses that KBTG serves. Additionally, this will help improve inclusivity in Southeast Asian LLM, resulting in models that not only understand the Thai language and cultural contexts but also those of neighboring countries. This effort will lead to the creation of diverse and high-quality datasets for training various models within the Southeast Asian Languages in One Network (SEA-LION). Everyone will be able to access the datasets and models of this project through open source, jointly advancing the development of the region's LLM ecosystem.

Dr. Leslie Teo, Senior Director of AI Products, AI Singapore, stated, "We are pleased to welcome KBTG as a partner and member of Project SEALD. KBTG's AI expertise, resources, and position as a leading tech organization in Thailand will serve to strengthen the Project SEALD network. We look forward to working with the KBTG team toward our shared vision of increasing Southeast Asian representation across the global LLM landscape."

Harsh Dhand, Head of Google Research & Core Partnerships, APAC, remarked, "Our work to advance understanding of regional languages is critical to bridge the digital divide, and to ensure that advancements in AI are equitable from the outset. Today, we are excited to welcome KBTG as a key partner for Project SEALD. Through this collaboration, our joint aim is to significantly enhance the availability of high-quality Thai language datasets for model training and evaluation, enabling the development of more robust and representative language models for the region."

ฝากข่าวประชาสัมพันธ์?

ติดต่อเราได้ที่ facebook.com/newswit