In the Internet era, recommender systems have become very important in daily life, and the combination of Generative Adversarial Networks (GANs) and recommender algorithms provides new opportunities for the development of this field. In order to solve the problem of computer education course resource recommendation, this paper combines the collaborative filtering recommendation algorithm and the sequence generation adversarial network to construct a generative adversarial network (MGFGAN) recommendation model with multi-dimensional gradient feedback, and designs a computer education course resource recommendation system with this as the core algorithm, and explores its role in improving the teaching effect. Compared with ItemPop, MF-BPR, and MGFGAN-I based on user interaction vectors, MGFGAN-A based on user attributes achieves optimal values in all recommendation performance indicators, and improves the performance of Precision@10, Recall@10, NDCG@10, MRR@10, and MRR@10 compared with MGFGAN-I, respectively. 0.0594, 0.0103, 0.0392, and 0.0829, respectively. Using the systematic clustering method, the results of the cluster analysis of the performance of the students in the experimental group using the recommender system of this paper and the control group not using the recommender system show that the experimental group of students achieved better results. This paper provides a methodological path for using generative AI to improve the teaching effectiveness of computer education.
