Chinese tech companies have been launching a multitude of artificial intelligence (AI) large language models (LLMs) but according to Baidu CEO, Robin Li Yanhong, this is “a huge waste of resources”. Speaking at the annual X-Lake Forum, Li called for the government to focus on developing applications that will bring ChatGPT-like services to market. He expressed concern over the fact that there have been 238 AI models launched in China as of October, up from 79 in June, yet there are very few successful AI applications that are familiar to the public. Li emphasized the need for a shift from developing foundation models to creating AI-native applications that leverage the capabilities of AI for economic growth.
The Chinese tech industry’s obsession with large language models has raised eyebrows. Li pointed out that the proliferation of these models is resulting in a lack of AI-native applications that could be useful to the public. He highlighted the need for scalable AI applications based on these models, rather than focusing solely on building bigger models. Li stated, “We need 100 million AI-native applications, but we don’t need 100 big models.” Large language models are deep-learning AI algorithms that can perform tasks such as recognition, summarisation, translation, prediction, and content generation using vast amounts of data.
Li called for industry policies to encourage the development of AI-native applications based on big models. He believes that doing so will create a prosperous AI ecosystem and drive economic growth. However, Li cautioned against recent efforts by some enterprises and cities in China to create their own AI foundation models from scratch. He argued that these models would lack “emergent abilities” due to their limited parameters and training data sets. The sophistication and capabilities of a model are measured by the number of parameters it has. OpenAI’s ChatGPT, for example, was trained on 175 billion parameters, while most Chinese LLMs have between 6 and 13 billion parameters.
While Baidu was the first major tech firm in China to launch its own AI chatbot, Ernie Bot, Li acknowledged that the best AI-native applications have yet to be developed. Baidu aims to incubate its own AI applications, such as the code-writing assistant Comate, but Li believes that there will be groundbreaking AI-native applications in the future. He drew a parallel to the era that gave birth to popular “mobile-native” apps like WeChat, Douyin, and Uber, asserting that excellent AI-native applications will emerge based on the big models.
In conclusion, the current frenzy over AI large language models in China has been deemed a waste of resources by Baidu’s CEO. The focus should shift towards developing AI-native applications that leverage the capabilities of these models to drive economic growth. While there is intense competition among Chinese tech firms to provide ChatGPT-like services, the real potential lies in the development of groundbreaking AI-native applications in the future.
Use the share button below if you liked it.