The Taiwanese venture capitalist and former president of Google China, Lee Kaifu, has just launched a new artificial intelligence start-up headquartered in Beijing.
The prime focus of the new company, named Lingyi Wanwu, will be large language models (LLM), a type of algorithm that can predict and generate content based on huge internet datasets. AI chatbots like ChatGPT and Baidu’s Ernie Bot are among the many possible applications of LLMs.
Lee believes Chinese tech firms developing their own LLMs from scratch, rather than relying on foreign open-source models, is key to the country becoming the global leader in the field. “We predict that once China has a truly native and high-quality large-scale model, the development of high quality and innovative applications will rapidly unfold, just like in the mobile internet and AI 1.0 eras,” Lee was quoted as saying.
Among the top talent poached for the new venture is Baidu’s former head of metaverse operations, Ma Jie, who left the search engine creator in May. Ma holds a 99% stake in Lingyi Wanwu, whilst Lee’s venture capital firm Sinovation holds the remaining 1%.
Lingyi Wanwu is entering a busy field, with tech giants Alibaba, Tencent, and Baidu all already working hard on LLMs. China reportedly has at least 79 LLMs with parameters exceeding 1 billion as of late May.
