Assessment of Outcomes and Benefits of National Multimodal Large Language Model Programme towards Singapore's National AI Strategy 2.0

Ms He Ting Ru asked the Minister for Digital Development and Information (a) whether the National Multimodal Large Language Model Programme launched in 2023 has met its stated objectives to date; (b) what outcome indicators and performance metrics are used to assess its progress; and (c) whether these outcomes are on track to deliver the intended benefits under Singapore's National AI Strategy 2.0.

Development of the two large language model (LLM) families under the National Multimodal Large Language Model Programme (NMLP) – Southeast Asian Languages in One Network (SEALION) and Multimodal Empathetic Reasoning and Learning in One Network (MERaLiON) – helped scientists and engineers in Singapore acquire capabilities to build and fine-tune models, produce datasets and customise models for our region's cultural and linguistic contexts.

Organisations, like the Monetary Authority of Singapore, and Indonesia's GoTo Group have used these LLM families to develop applications that better serve users in Singapore and Southeast Asia. The development teams for SEALION and MERaLION also collaborate with companies, such as Google Research and Microsoft Singapore, to build better models. These outcomes are in line with our National Artificial Intelligence Strategy 2.0 objectives.