At present, the global artificial intelligence has entered a critical period of in-depth application of large models and rising strategic demand, how can China seize industrial opportunities and improve its international competitiveness? At this year's National People's Congress and the National People's Congress, Liu Qingfeng, chairman of iFLYTEK, submitted a proposal on systematically accelerating the development of China's general artificial intelligence, from multiple dimensions such as policy support, independent and controllable computing power ecological construction, data open sharing, evaluation standard formulation, source innovation, talent training, law formulation and ethics and humanities, to enhance the strength of China's general artificial intelligence industry in the general base platform, and to build China's comparative advantages in industry application and value creation.
Liu Qingfeng, deputy to the National People's Congress and chairman of iFLYTEK.
He believes that under the guidance of the "New Generation Artificial Intelligence Development Plan" issued in 2017, China has a very solid technical reserve and an established team in the field of cognitive intelligence.
At present, China has formed an international leading comparative advantage in the fields of speech model and medical model, but with the release of GPT-5, there is still room for improvement in computing power, data, model training and other aspects of China's large model.
At present, the core and main battlefield of general artificial intelligence is the competition in the ability of general base. Focusing on the independent and controllable base model, focusing on resources at the national level to accelerate catch-up, it is expected to achieve transcendence in typical industry fields. Liu Qingfeng suggested that, first of all, we should increase and maintain continuous investment in the general large model base, support research and development, and formulate relevant policies to promote the application of large models in the fields of industry and people's livelihood, so as to form a good situation of mutual promotion of "base large model + industry application".
Secondly, accelerate the formation of an independent and controllable industrial ecology around the domestic model. For example, support the opening of domestic large models to developers, carry out the construction of large model evaluation system and open source community, and reduce the cost of R&D and use; Support the application of industry models to automobiles, home appliances, service robots and other industries, and promote the vigorous development of China's independent and controllable large model industry ecology with "artificial intelligence +".
At the same time, we will promote the openness and sharing of high-quality training data at the national level, and build a multi-source and multi-modal national data resource aggregation platform including national public data resources, high-quality e-books, high-quality audio data, multi-channel industry application data, and Internet open source data resources.
Finally, a more objective, fair and credible evaluation method should be formulated to accelerate the application of large models in the industry. Adhere to the systematic innovation of core technologies at the source, and do a good job in the strategic and forward-looking basic research field; Accelerate the promotion of large models to empower the whole school period, and explore new models for the cultivation of top-notch innovative talents and application-oriented talents in artificial intelligence in China; Accelerate the formulation and review of laws and regulations related to general artificial intelligence technology, and improve the reliability and standardization of general artificial intelligence technology; Set up soft topics to conduct ethical and humanistic research related to general artificial intelligence, and deal with the social risks, ethical challenges and changes in human civilization that may be brought about by general artificial intelligence technology.
Hui Shang "full**.
Text |Hui Shang" full ** reporter Shao Meng.
Figure |Courtesy of the interviewee.
Editor-in-Charge |Horse Garden.
Crafting |Wang Yulu.
First instance |Qi Fei.
Second instance |Liang Shuang.
Third Instance |Bao Xiaochun.