US firm's AI tech ban set to inspire homegrown innovation

2024-08-08 13:26
SHI YU/CHINA DAILY

US-based research company OpenAI's recent move to block access in China and other countries to the technology used to build artificial intelligence products is likely to backfire and boost development of the domestic industry, experts said.

The ban will motivate Chinese large language model companies to enhance their own innovation capacities and encourage more AI startups to opt for homegrown LLMs to avoid potential geopolitical risks, they said.

"The suspension of OpenAI's services in China will accelerate the development of the domestic LLM industry," said Zhou Hongyi, founder of the cybersecurity company 360 Security Group. "It might not be a bad thing after all," he said in a video posted on social media platform Sina Weibo.

Industry insiders also encouraged Chinese LLM companies to invest more in improving computing power and algorithms, and accumulate more high-quality training data to gain an edge in the global AI race and narrow the gap with the United States on the latest tech frontier.

LLMs are AI models fed with huge amounts of text data for use in a variety of tasks, ranging from natural language processing to machine translation. It is the key technology underpinning OpenAI's ChatGPT, which has taken the world by storm since late 2022.

Last month, the US company announced the restrictions on its application programming interface, or API, effective from July 9. API is a set of rules or protocols, that enables software applications to communicate with each other to exchange data, features and functionality. It allows third-party developers to integrate AI models into their applications.

Although ChatGPT is not available on the Chinese mainland, some Chinese startups have been able to access OpenAI's API platform and use it to build their own applications.

The San Francisco-headquartered firm's decision also came on the heels of a draft rule released by the US Department of the Treasury on June 22 that would restrict and monitor US companies' and individuals' investments in China for AI, chips and quantum computing.

Luring new users

In response to OpenAI's announcement, a slew of Chinese technology companies are scrambling to lure OpenAI users to their platforms by offering migration options and free tokens, the smallest unit text data can be broken down into for an AI model to process.

Among them is tech heavyweight Baidu Inc, which has rolled out an inclusive program for users to have a smooth and free transition to its LLM Ernie platform.

The company is also providing free AI model fine-tuning and expert guidance on its flagship Ernie model, along with 50 million free tokens, which developers can use to query the AI-powered chatbot.

Alibaba Cloud, the cloud computing unit of Alibaba Group, is offering 22 million free tokens and migration services for OpenAI's API users through its own AI platform.

The company's Qwen-plus model is priced significantly lower than GPT-4, it said. Tongyi Qianwen 2.5, the latest version of its LLM, has caught up with the GPT-4 model in multiple capacities, including text understanding and generation, knowledge quizzes, dialogues and life advice, in the context of the Chinese language.

Other companies offering incentives to migrate include startup SenseTime and Baichuan Intelligent Technology, which is backed by Alibaba and Tencent Holdings.

Zhipu AI, a Beijing-based startup and one of the four new "AI tigers" of China, has announced a special migration program that includes training and consulting services, and 150 million free tokens. Unicorn company 01.AI, founded by renowned AI expert Kai-Fu Lee, also announced in late June a half-price replacement plan for users to migrate to the company's Yi series large models.