Chinese AI Firms Boost Local Language Models in Global South

Chinese AI firms are expanding their cooperation in South-South regions, focusing on local language models. In Kenya, a Chinese tech company supports a digital human at Safaricom, demonstrating services and interpreting data. What's interesting is that these firms are partnering with host nations to build large language models specifically for local languages and industrial needs. This addresses a gap, as leading global models often prioritize widely spoken languages, leaving other communities underserved. For example, Chinese tech company iFlytek has developed its Spark large language model. This model supports speech recognition and simultaneous interpretation for over 130 languages. They've also created the Spark ASEAN Multilingual Large Model Base, covering 10 languages like Malay, Indonesian, Vietnamese, and Thai. This base performs comparably to top international models but with smaller parameter sizes, allowing for efficient deployment in ASEAN countries. The bottom line is that Chinese AI companies are providing customized AI solutions for developing countries, recognizing the strong demand for models adapted to local languages. This matters because it helps bridge the technological gap for communities with less-used languages.