Kai-Fu Lee, a renowned computer scientist and author of “AI Superpowers,” has embarked on a new venture with 01.AI, aiming to create a homegrown large language model (LLM) tailored for the Chinese market. This endeavour puts Lee in competition with other notable Chinese tech leaders striving to establish China’s equivalent of OpenAI. The motivation behind this initiative lies in the necessity arising from the unavailability of OpenAI and Google products in China, prompting the need for homegrown solutions.
In just seven months since its inception, 01.AI has released its first model, the open-source Yi-34B. This bilingual LLM, with 34 billion parameters, has already gained recognition as it outperformed larger models in a ranking by Hugging Face. The startup’s ability to quickly develop and train models can be attributed to successful fundraising, securing substantial financial backing from investors like Sinovation Ventures and Alibaba Cloud.
01.AI has grown rapidly, now employing over 100 experts in LLM technology, including talents from multinational and Chinese tech firms. The team behind Yi-34B includes individuals with experience from Microsoft Research Asia, Google’s Bard, and TensorFlow, as well as notable figures from the AI industry. With a wealth of connections and a history of nurturing talent, Kai-Fu Lee is well-positioned to lead this ambitious project.
Recognizing the cost-intensive nature of building LLMs, 01.AI plans for monetization from the outset. While some models will remain open source, the company aims to develop proprietary models that can serve as the foundation for various commercial products. This strategy allows the company to invest in the necessary computing power, including GPUs, which are critical for LLM development.
Anticipating potential U.S. sanctions on high-end AI chips, 01.AI proactively stockpiled GPUs, ensuring a stable supply for the coming months. The company’s path to monetization hinges on finding the right product-market fit for its expensive artificial intelligence (AI) models. In China, top-notch LLM scientists are scarce, but there is a wealth of product talent, particularly in the mobile internet ecosystem.
Further Reading on AI vs Layoff: Elon Musk Predicts AI’s Potential to Render Jobs Obsolete
While details of the services in the pipeline remain undisclosed, Kai-Fu Lee hinted at experiments in productivity and social directions. He expressed the desire to release an app within the calendar year. Ultimately, 01.AI’s goal is to become an ecosystem that enables external developers to easily build applications, emphasising the importance of making application development accessible and fostering compelling applications.
Kai-Fu Lee’s 01.AI venture holds significant promise in the evolving field of LLMs, aiming to bridge the technological gap in the Chinese market while contributing to the global AI ecosystem. Time will reveal the success of this ambitious endeavour.