With the rapid advancement of artificial intelligence (AI) technology, academic search processes are being significantly transformed. Traditional academic search methods are often time-consuming and labor-intensive, requiring substantial effort from researchers for background research and literature reviews. AI technology offers a solution by enhancing the efficiency and accuracy of literature retrieval through natural language processing, machine learning, and deep learning. This study aims to understand user needs and expectations for AI-based academic search tools through a survey. The results indicate that users highly value features such as summary and content extraction for papers/theses, automatic translation, and text-data metadata analysis. Factor analysis categorizes essential features into four primary factors: non-text material analysis, customized searching, text material analysis, and advanced features. These findings highlight the potential of AI to revolutionize academic search, making it more efficient and comprehensive, and provide insights for the development of user-centered AI academic search tools.