Indias AI build-out a reference for Taiwans approach

TAKING A CUE: Looking at India’s approach to their vest web of languages, experts say that could be a useful reference as Taiwan develops its own local AI

• Staff writer, with CNA

Taiwanese experts attending the 4th artificial intelligence (AI) Impact Summit in New Delhi said they were struck by India’s development of AI models capable of handling dozens of dialects, and described India’s approach as a valuable reference for Taiwan’s own “sovereign AI” efforts.

The summit, held from Monday to Friday last week, brought together global tech industry leaders and political figures from multiple countries, with Taiwan also sending representatives.

Liu Ching-yi (劉靜怡), Taiwan AI Center of Excellence vice chair and distinguished professor at National Taiwan University’s Graduate Institute of National Development, said that one of the summit’s central themes was how to make AI more equal, inclusive and diverse for people around the world.

“With government support, Indian companies are working on sovereign AI to protect cultural and linguistic diversity,” Liu said, adding that “many companies” have launched mature language models that can be compatible with 20 to 30 dialects and have developed them into products.

“Taiwan should also strive to develop in a similar direction, so that while innovating in AI development, it can also maintain linguistic and cultural diversity and ensure the basic rights of people,” Liu said.

“Taiwan should also think about how to make good use of its advantages in chips to gain more of a say in international AI cooperation and competition,” she added.

Taiwan AI Academy CEO Chen Ling-jyh (陳伶志) said that AI could be used to “develop some government services,” since Mandarin, Hakka and Hoklo (commonly known as Taiwanese), as well as indigenous languages, are all spoken in Taiwan.

Hsia Yun (夏耘), a researcher at Taiwan’s National Institute of Cyber Security’s Center of International Collaboration and Cybersecurity Governance, said that “international cooperation is continuing” in the fields of AI safety and governance.

“Taiwan can have more influence and contribute more on AI issues, and I hope that in the future Taiwan can establish substantive cooperation with various countries and participate more deeply in discussions on AI governance,” Hsia said.

During the summit, Indian Prime Minister Narendra Modi met with French President Emmanuel Macron, Brazilian President Luiz Inacio Lula da Silva and other leaders in an effort to strengthen global partnerships in AI.

OpenAI CEO Sam Altman, Google CEO Sundar Pichai, Meta chief AI officer Alexandr Wang, Google DeepMind CEO Demis Hassabis and Anthropic CEO Dario Amodei also attended the summit.