Without local languages, ‘AI is essentially useless’: Hong Kong’s Votee AI is taking on English and Mandarin’s AI dominance with a Cantonese model
Votee is one of a growing number of companies that are trying to tackle AI’s neglect of "low-resource" languages.
The AI boom is primarily driven by English and Mandarin Chinese. Major AI developers such as OpenAI, Anthropic, DeepSeek, Moonshot AI, and Z.ai are based in the U.S. or mainland China, leading to these languages dominating AI performance. This leaves other languages, spoken by hundreds of millions of people, underserved. Hong Kong-based startup Votee AI aims to change this by creating Cantonese and other neglected language models.
CEO Pak-Sun Ting emphasizes that AI currently neglects many languages, and without support, it is essentially useless. Cantonese grammar, vocabulary, and colloquial usage differ significantly from Mandarin, even though over 80 million people speak it. Standardized written data for colloquial Cantonese is limited, and mainstream AI models struggle with cultural knowledge.
Votee trims open-source models from Meta and Alibaba, retraining them on Cantonese data, resulting in models costing around $250,000 and using 500 million to 1 billion tokens. These models can understand and reason in Cantonese while being significantly cheaper to train than English-language counterparts. As a for-profit company, Votee serves governments and corporations needing local-language AI.
Ting envisions a future where countries own their AI supply chains, but acknowledges this is challenging.
Written by urgent.news from Fortune's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
