The core data of this article: the sources of AI core technologies for Chinese enterprises in 2020, the growth of corporate income driven by the transformation and application of AI technologies in China, the current status of the application technology direction of enterprise AI projects, the application track and segment ratio of AI technologies.
1.Computer vision is currently the main application technology
2. Companies are starting to focus on AI self-research capabilities
From the perspective of the enterprises that have applied AI projects, in the analysis of the sources of enterprises to obtain AI core technology, 61% of the enterprises chose the independent research and development path, and about 40% each chose to call the third-party platform AI technology through AIPD/SDK to choose the form of industry-university-research cooperation and development; 31.7% of the enterprises used open source technology; 26.8% of the enterprises chose to entrust third-party companies to provide solutions. The percentage of enterprises that choose to commission external professional AI companies to develop is 17.1 percent.
The typical application scenarios of various AI technologies applied by enterprises are gradually becoming clearer. The application scenario of computer vision is mainly concentrated on security monitoring and interactive large screen speech recognition technology, while the selection rate of AI customer service system and intelligent voice assistant both exceed 70%. Natural language processing is mainly used for intelligent review and report generation, knowledge mapping is mainly applied to assisted decision making and intelligent diagnosis scenarios, and the application scenarios of machine language learning are more distributed, with relatively high selection rates for predictive models and intelligent risk control. The selection rate of predictive models and intelligent risk control is relatively high.
Computer vision and machine learning, both included in the major applied technologies for the next three years, dropped to second and third place with 48.8% and 46.3% respectively, while deep learning took first place with 53.7%.