Artificial Intelligence (AI) is transforming industries and societies worldwide, but its true potential can only be realised when it is accessible and relevant to diverse linguistic and cultural contexts. In Asia, a region characterised by its diverse range of languages and cultures, developing AI in local languages is not just a technological challenge but a socio-economic imperative. Asia is home to over 2300 languages, with millions of people speaking languages that are often underrepresented in global AI developments.
Developing AI in local languages is not just about technological advancements; it’s about inclusivity. It ensures that everyone can benefit from AI applications in healthcare, education, and public services regardless of their language. For instance, AI-powered educational tools in local languages can bridge educational gaps in rural areas, providing quality education to children who might otherwise be left behind. This emphasis on inclusivity makes AI development more meaningful and impactful.
AI in Local Languages Can Drive Economic Growth
AI systems and tools in local languages can drive economic growth by enabling businesses to reach wider audiences. e-Commerce platforms, customer service bots, and marketing tools that understand and communicate in local languages can significantly enhance customer engagement and satisfaction. According to Omdia report, the total software revenue for generative AI in Asia and Oceania is expected to grow to $18.3 billion by 2024, with a compound annual growth rate (CAGR) of 53%. This growth is fueled by the development of AI solutions tailored to local markets.
Why is it Difficult to Develop AI Systems in Local Languages?
Among the AI biases, data scarcity and linguistic complexity are crucial to be addressed. Many local languages lack extensive digital corpora, making it difficult to train robust AI models. Initiatives like Project SEALD (Southeast Asian Languages in One Network Data) are addressing this by collecting and curating multilingual datasets for languages such as Indonesian, Thai, Tamil, Filipino, and Burmese.
Local languages often have complex grammatical structures, diverse dialects, and unique cultural contexts that pose challenges for AI development. For example, tonal languages like Thai and Vietnamese require sophisticated models to interpret and generate speech accurately. However, addressing these complexities is not a task for AI researchers and linguists alone. It requires the active participation and insights of local communities, empowering them to shape the future of AI in their languages. The significant need for computational resources and expertise is also a big challenge in developing AI in local languages since many Asian countries have limited access to them.
Recent Advancement and Initiatives
AI Singapore’s SEA-LION project is a pioneering effort to create LLMs that are culturally and linguistically representative of Southeast Asia. The project involves building a diverse and high-quality data corpus to support the training of these models. Project SEALD, a collaboration with Google Research, enhances datasets for languages spoken across Southeast Asia and develops tools for translocalisation and translation.
Another significant initiative is Singapore’s National Multimodal LLM Programme, launched in 2023. This programme is a crucial step towards developing AI in local languages. Funded by the National Research Foundation, this $70 million initiative aims to create AI models that incorporate the diverse cultures and languages of Southeast Asia. The programme also focuses on building AI talent and fostering a thriving AI industry in the region.
Japan has also committed to helping Southeast Asian countries train large language models in their local languages. This support includes providing technical expertise and resources to develop AI solutions that cater to the region’s linguistic diversity. Such international collaborations are crucial for advancing AI in local languages and ensuring that the benefits of AI are widely distributed.
The Road Ahead
Developing AI in local languages is complex but an essential endeavour for Asia’s digital future. By addressing challenges such as data scarcity, linguistic complexity, and resource constraints and leveraging regional collaborations and investments, Asia can harness the full potential of AI. The ongoing initiatives and advancements in this field are promising steps towards a more inclusive, culturally rich, and economically vibrant future. As AI continues to evolve, it is imperative that efforts to develop AI in local languages are sustained and expanded. This will not only ensure that technological advancements are accessible to all but also preserve and promote Asia’s rich linguistic and cultural heritage.