Latest news with #IntelekLuhurMalaysiaUntukmu
New Straits Times
3 days ago
- Business
- New Straits Times
Gobind: Malaysian-built ILMU AI no threat to jobs
KUALA LUMPUR: The launch of Malaysia's first locally trained large language model (LLM), ILMU, will not threaten jobs as the technology itself is not the main disruptor, says Digital Minister Gobind Singh Deo. He said that an LLM alone would not displace workers. "It's an LLM. Your question is more on AI in general. Of course, AI will have an impact on jobs. "That is something we have said from the start. But it will also create new jobs," he said after the launch of the model, developed by YTL. He was responding to reporters after the launch of Malaysia's sovereign AI platform, ILMU — short for Intelek Luhur Malaysia Untukmu (Excellent Malaysian Intelligence for You). The minister said the government was already taking steps to mitigate AI's impact on employment. "These jobs that are going to be impacted, we are already in the process of looking at the data. "And we want to make sure that in those key areas, or for those key jobs that are going to be affected, we are able to upskill and reskill the workers. "So when that impact occurs, the workers affected are already transitioned… they are ready to continue working based on the reskilling and upskilling they have received," he said. He described ILMU as a "Malaysian LLM" trained in multiple languages used in the country, incorporating local cultural and linguistic nuances. "ILMU is an LLM trained in all aspects of Malay. It knows the language… its training is based on Malaysia's conditions. "So apart from language, it is also trained in Malaysian ways," he said, adding that YTL AI Labs, ILMU's developer, would issue a detailed statement on its features. The model was launched as part of efforts to build AI systems tailored to Malaysia's needs. Gobind Singh earlier delivered the opening address at the Asean AI Malaysia Summit 2025's opening ceremony. The launch of ILMU marked the start of an inaugural gathering to chart the direction of AI in Malaysia, particularly in governance. The two-day summit at the Malaysia International Trade and Exhibition Centre (Mitec) is organised by the Digital Ministry. The opening ceremony also saw Chief Secretary to the Government Tan Sri Shamsul Azri Bakar, Asean secretary-general Dr Kao Kim Hourn, and Asean delegates, alongside global experts and industry players, in attendance. Malaysia's sovereign AI platform, ILMU, is a homegrown LLM developed by YTL AI Labs. Built to understand and process text, voice, and images, it is designed with a strong emphasis on local context, ethics, and safety. The first iteration, ILMU 0.1, was unveiled in December 2024 and outperformed leading global LLMs in Malay-language benchmarks, even passing Malaysia's PT3 and SPM national exams with top grades.
The Sun
3 days ago
- Business
- The Sun
Malaysia launches ILMU: First locally built AI language model
KUALA LUMPUR: Prime Minister Datuk Seri Anwar Ibrahim (pic) today launched Intelek Luhur Malaysia Untukmu (ILMU) — the nation's first 100% locally developed multimodal large language model (LLM) — set for early access on September 16 via ILMUchat. Speaking at the ASEAN AI Malaysia Summit 2025 (AAIMS25), Anwar said the goal is to position Malaysia as an 'AI nation' that uses technology to improve governance, drive innovation and uplift communities. 'This is a generational opportunity. AI can bridge divides, empower small businesses, strengthen public services and expand access to healthcare, education and justice for all,' he said. Developed by YTL AI Labs in collboration with Universiti Malaya, ILMU integrates local languages and culture with world-class performance. Digital Minister Gobind Singh Deo called it a 'groundbreaking Malaysian innovation' that safeguards linguistic and cultural identity while delivering global-standard AI capabilities. Fluent in Bahasa Melayu, Malaysian English (Manglish) and dialects such as Kecek Kelate, ILMU is built on YTL AI Cloud for enterprise-grade performance and local data residency. It reportedly matches or outperforms GPT-4o, DeepSeek and Llama 3.1 in key benchmarks, and leads globally in Malay language understanding. The model can process and generate text, voice and images, enabling applications across multiple sectors.
