Baidu CEO Warns Potential Resources Concession In China's Big Language Model Development

JAKARTA - On Wednesday, November 15, CEO of one of China's leading artificial intelligence companies, Baidu, warned that craze in China's development of major language models could lead to resource waste. Robin Li, CEO of Baidu, stated that Baidu needed to focus on developing practical applications instead.

The comments were made by Li in an industrial forum in Shenzhen, as market concerns grew over potential industrial shocks as companies developing large language models have not found a viable business model.

Since the release of the ChatGPT chatbot by OpenAI last year, generative artificial intelligence has attracted great attention in China, where established companies and startups have also taken part in this field.

"I have observed a phenomenon (in China) where many industries, companies, even cities, buy hardware, store chips, (and) build computing centers to train their large models from scratch," Li said.

"A large language model itself is a fundamental basis like an operating system, but in the end developers need to rely on a limited number of large models to develop various original applications. Therefore, constantly developing a large fundamental model is a huge waste of social resources," Li said., quoted by VOI from Reuters.

Li said that although there are many major models in China, the artificial intelligence applications developed based on these models are still very few.

As of October, 238 major language models have been released, increasing from just 79 in June, according to Li citing third-party reports.

The big Baidu language model, called Ernie, opened for public use in August, joined other products that have received government approval for release. Last month, Baidu introduced the latest version of its generative artificial intelligence model, Ernie 4.0, whose first version was launched in March.