Chen Li
ANBOUND advocates that China should not develop large AI systems, touted as “big AI”, but instead focus on smaller ones. When discussing the rationale behind such analysis and proposition, the cost factor is one crucial point.
At the end of 2022, ChatGPT emerged and signaled the age of the rapid development of AI globally. From the United States to China, large language models similar to ChatGPT have mushroomed, becoming a valuable resource for various technology companies to showcase their strengths and attract capital. Public information shows that by the end of March 2024, nearly 120 large models have been registered and launched in mainland China. The large models that have completed filing for generative AI services include Baidu’s Ernie Bot, Alibaba’s Tongyi Qianwen, Huawei’s Cloud Pangu, Tencent’s Hunyuan, OPPO’s AndesGPT, Vivio’s BlueLM and others. If including large models that have not been registered, the total number of AI large models in China reached 238 in 2023.
From the perspective of speed and quantity of research and development, China does not seem to lag far behind the world’s leading level in terms of AI large models. At many large model launch events in the countries, major tech companies emphasize the superiority of their large model performance, often claiming to be “comparable to” or even “ahead of” ChatGPT. For example, at the upgrade release conference of the StarFire Cognitive Large Model V1.5 held by iFlytek in June last year, there were several mentions of “StarFire is just a step away from ChatGPT” and “will surpass ChatGPT before October this year”.
No comments:
Post a Comment