Learning emerging and popular AI technologies such as machine learning and deep learning through online learning videos has become an important way for people to quickly keep up with AI related technologies. However, there is a lack of research on the factors that affect the communication effectiveness of these knowledge learning videos. This study constructs a theoretical framework for the influencing factors of the communication effectiveness of AI knowledge learning videos based on Lasswell’s 5W model and DeFleur’s interaction process theory. Taking the Bilibili platform as the research object, from the aspects of online video characteristics (Provider) and users’ emotional attitudes (User), with “machine learning” and “deep learning” as keywords, this study collects the comment and behavior data of relevant online learning videos and analyzes them by means of descriptive statistics, text sentiment analysis, and topic clustering. The research finds that the account characteristics of the communication subject (whether it has obtained official platform certification), the communication content (whether it incorporates popular hot topics), and the communication interaction (whether it gains users’ emotional recognition) are the key factors affecting the communication effectiveness of online AI knowledge learning videos. Based on this, multiple strategies are proposed to improve the communication effectiveness of AI knowledge learning videos. This study is of great significance for enhancing the communication effectiveness of AI knowledge learning videos and promoting the high-quality development of knowledge learning platforms.