China News Service, Beijing, March 6 (Xu Xueying) "We must face up to the gap, focus on the 'main battlefield' of autonomous and controllable base large models, focus resources from the national level to accelerate catch-up, and at the same time systematically build a general artificial intelligence ecosystem and applications. Create comprehensive advantages." Liu Qingfeng, deputy to the National People's Congress and chairman of iFlytek, said.
This year’s government work report pointed out that we will deepen the research and development applications of big data, artificial intelligence, etc., carry out the “artificial intelligence +” action, and create an internationally competitive digital industry cluster.
In recent years, general artificial intelligence has become a global craze.
Some people in the industry believe that the wisdom emerging capabilities presented by large models will completely change the industrial form and competition landscape.
"Under the guidance of the "New Generation Artificial Intelligence Development Plan" issued in 2017, China has very solid technical reserves and an established team in the field of cognitive intelligence, and is expected to become the second pole of global wisdom." Liu Qingfeng said.
How far are domestic large models from the world's highest level?
Liu Qingfeng took the iFlytek Spark model, China's first large-scale public model based on national industrial computing power training, as an example. "It is expected to reach the current best level of GPT4/4V within 6 months." However, as GPT- 5, this gap may be stretched to more than a year. If resources are organized to catch up in terms of computing power, data, model training, etc., this gap is expected to be tied to a considerable level within 1-2 years. At the same time, China It has also formed internationally leading comparative advantages in fields such as speech large models and medical large models.
"We are confident that there will be no modern backwardness gap on the universal large model base. On this basis, we will polish it based on industry scenarios and data, and it is expected to achieve transcendence in typical industry fields." Liu Qingfeng believes that while catching up, combined with Brand-new technological development, competitive landscape, industrial empowerment, etc., it is necessary to formulate systematic plans based on the new situation.
He suggested that on the basis of the 2017 "New Generation Artificial Intelligence Development Plan", the design should be aimed at the shortcomings that need to be addressed in the development of China's general artificial intelligence, and should be designed around the construction of an independent and controllable computing power ecosystem, the open sharing of high-quality data, Systematically formulate a national general artificial intelligence development plan from such dimensions as the formulation of scientific evaluation standards, forward-looking research and development of source technologies, talent training, legal formulation, ethics and humanities.
Focusing on talent cultivation and introduction, Liu Qingfeng suggested accelerating the promotion of large-scale models to empower all academic levels, and accelerating the exploration of the cultivation of China's top innovative artificial intelligence talents with new mechanisms, such as the establishment of the National Artificial Intelligence Academy.
(over)