INDONESIAN e-commerce platform Tokopedia and Singapore IT player NCS are among the companies that are testing an artificial intelligence (AI) large language model (LLM) being developed in Singapore.

The LLM is being developed by AI Singapore, a national AI programme. LLMs are AI models that can understand and generate human-like speech or conversation.

Dubbed Sea-Lion – short for South-east Asian Languages in One Network – it will be trained to understand and generate output that incorporates the region’s diverse languages and cultures.

The model’s use cases currently include e-commerce, legal, information technology and customer-service functions, said Leslie Teo, senior director of AI products at AI Singapore, at a media briefing on Wednesday (Jan 24).

Tokopedia, for example, uses Sea-Lion to generate product descriptions in other South-east Asian languages.

NCS uses the model to translate content used in its legal compliance process into Bahasa Indonesia and Thai. The AI models it had been using were built on LLMs in English, said Dr Teo.

A NEWSLETTER FOR YOU

Friday, 8.30 am

Asean Business

Business insights centering on South-east Asia's fast-growing economies.

Sign Up

The model can also be used in customer-service chatbots to capture nuances in South-east Asian culture, and to summarise information in regional languages.

Sea-Lion has started off with the more commonly used languages in the region, such as Bahasa Indonesia, Malay, Thai and Vietnamese; it will eventually expand to include other languages, including Lao and the Myanmar language.

The model is open-source, meaning it is publicly available for companies and developers to use as a foundation model to be customised to their own needs.

Sea-Lion is part of a S$70 million AI initiative in Singapore that was unveiled last December. The bulk of the funding, from the National Research Foundation, will go towards recruiting talent and obtaining computing infrastructure, said Dr Teo.

AI Singapore works on the model with partners such as Amazon Web Services for infrastructure and Google Research for data; Sea-Lion also taps SeaCrowd, a collaborative project that collects datasets about South-east Asian languages and cultures.