This paper explores new approaches to the application and creation of digital media in the context of the rapid development of artificial intelligence technology, with a particular focus on virtual reality interaction. The study utilizes the spatial distance of the enclosing spheres in the hybrid hierarchical enclosing box and the characteristic of non-intersection between the enclosing boxes for collision detection, and the collision information returned by the rays is analyzed in depth. Meanwhile, the intersection test method is used to improve the efficiency of collision detection. And a subjective evaluation of 10 users’ experience in virtual reality art interaction was conducted. The results showed that the system was highly rated in accuracy (average score of 8.39), but rated lower in simplicity (average score of 4.28). There was a positive relationship between the user’s eye gaze duration and the number of interruptions and misoperations in art interactions. At an eye gaze duration of 0.2 seconds, half of the users did not experience interaction interruptions; when the gaze duration increased to 1.2 seconds, all users had no interruptions. This paper provides a practical reference case for using AI technology in digital media applications and creation, especially in enhancing the accuracy and efficiency of virtual reality interactions.