Personalized recommendation affects the consumption behavior by collecting and analyzing theuser's behavior information, predicting the user's interest and preferences and making recommendations, therebygenerating economic benefits. Personalized recommendation has gone through the development of statistics-based, content-based, collaborative filtering, social network. -based and hybrid recommendation. Although it hasachieved certain results, it is still not satisfactory. With the advent of the era of artificial intelligence, multipledisciplines and fields provide new approach for personalized recommendation. This article first reviews the mainmethods about personalized recommendation, and analyses problems and potential needs. Then, based on thetheoretical models of management and psychology, it discusses that personalized recommendations in the eraof artificial intelligence need to be people-oriented and focus on user's psychological state and proposes severalways to improve the effectiveness of personalized recommendations.