With the development of artificial intelligence technology, the field of cultural communication is also gradually becoming widely used in artificial intelligence-related technologies. Aiming at the phenomenon of lagging development of national sports culture, this paper proposes a technical framework for optimizing the communication mode of national sports culture development with data collection and accurate information-pushing technology as the core. In order to collect data, the social network is transformed into a network graph model, and the Markov chain Monte Carlo method is used to ensure random and probabilistic sampling. The accuracy of prediction results is improved by establishing user profiles, using support vector machines, and combining the real-time computing framework to develop an accurate information delivery model. In the analysis of the development level of China’s wushu culture communication, it can be learned that the total reading volume of wushu culture-related topics on the microblogging platform totaled about 4.81 billion, the total discussion volume was 431,000, and the video playback volume of the related issues on the Jieyin platform reached 534 million times. The average value of the overall evaluation of the soft power of wushu culture is 3.83, which is close to the “good” evaluation level. Combined with intelligent technologies such as data collection and precise information pushing, the communication of Wushu culture has achieved better results.