The Science, Technology and Innovation Ministry (MOSTI) is developing a Bahasa Malaysia Large Language Model (LLM) to support Malaysia’s AI ecosystem while ensuring data security and alignment with local values.
Deputy Minister Datuk Mohammad Yusof Apdal highlighted that this initiative aims to reduce reliance on foreign technology and enhance decision-making, automation, and research across sectors by adapting global AI advancements to local needs, according to The Malaysian Reserve.
Developing an LLM is costly and requires specialised resources, often involving cloud-based high-performance computing.
To manage costs, MOSTI is exploring partnerships with cloud service providers like Microsoft Azure, OpenAI, and AWS (Amazon Web Services), though concerns about data security persist due to potential risks of sensitive information leaks.
Mohammad Yusof added that collaboration with Phison Technology via the aiDAPTIV+ platform has shown promise in lowering costs and improving data security by enabling local customisation and optimisation of AI models.
This initiative is guided by the AI Governance and Ethics (AIGE) framework introduced in September 2024, ensuring responsible AI development that respects Malaysian culture and values.
MOSTI is also working with other ministries and agencies to establish a comprehensive framework for the LLM initiative.
What is a LLM?
LLMs are machine learning models that can understand and produce human language text. They operate by analysing vast datasets of language.
LLMs are trained on vast amounts of data collected from the internet, encompassing thousands or even millions of gigabytes of text.
However, the quality of these data samples significantly affects the LLMs’ ability to learn natural language effectively.
Examples of real-world LLMs include ChatGPT (from OpenAI), Gemini (Google), and Copilot (Microsoft).