A Review of Generative AI in Recommendation Systems

Abstract

The development of Gen AI has been largely enabled by traditional machine learning methods, which have allowed
for the creation of increasingly complex models. Gen AI's use in recommendation systems is contrasted with that of
more conventional AI techniques, as well as the applications of both types of AI in other domains, in this review. By
fixing the problems with traditional machine learning methods, generative AI has revolutionized recommendation
systems. Focusing on GANs and VAEs, this paper analyzes the use of Gen AI and traditional AI in RS across
different domains. Low data, cold start, and suggestion variety are some of the issues that these generative models
highlight as strengths and weaknesses. Further study might be hindered by them as well. This research summarizes
current efforts to illustrate how Gen AI improves RS performance, follows important trends, and answers pertinent
new questions about ethics and dependability. Hybrid techniques are highlighted by the findings, which bode well
for future advancements of the effective and versatile RSs. The report ends by stating that further research should be
conducted.