Making LLMs more inclusive, diverse with external collaborations: AI Singapore

Singapore’s agency for AI development hopes its SEA-LION will add the necessary attributes to open source generative AI large language models

In the case study of ensuring that AI is inclusive and representative of global diversity, Singapore’s AI Singapore (AISG) is researching the creation of large language models (LLMs) for South-east Asian Languages In One Network (SEA-LION).

The aim is to address what AISG refers to as “gaps” in the representation of South-east Asian languages in the global LLM landscape. With the open source SEA-LION included for use around the world, GenAI language tools will be “more representative of all languages and populations.”

On 10 Sep 2024, AISG announced the signing of a Memorandum of Understanding with Sony Research and its subsidiary, Sony AI. Under the agreement, AISG and Sony Research will explore testing and feedback of the SEA-LION model, particularly for Tamil and other regional languages, and share best practices on LLM development and research methodologies.

According to AISG’s, Senior Director of AI Products, Leslie Teo, “the integration of the SEA-LION model, with its Tamil language capabilities, holds great potential to boost the performance of new solutions. We are particularly eager to contribute to the testing and refinement of the SEA-LION models for Tamil and other South-east Asian languages, while also sharing our expertise and best practices in LLM development. We look forward to seeing how this collaboration will drive innovation in multilingual AI technologies.”

Noted Sony Research’s president Hiroaki Kitano, who is currently a member of Singapore’s Advisory Council on the Ethical Use of AI and Data, the Infocomm Media Development Authority, the Singapore Economic Development Board, and the National Research Foundation: “Access to LLMs that address the global landscape of language and culture has been a barrier to driving research and developing new technologies that are representative and equitable for the global populations we serve,” adding that “linguistic diversity underscores the importance of ensuring AI models and tools are designed to support the needs of all populations around the world. We look forward to our collaboration with AISG and the potential to make AI work for everyone.”