Using information technology tools for academic help-seeking among college students has become a popular trend. In the evolutionary process between Generative Artificial Intelligence (GenAI) and traditional search engines, when students face academic challenges, do they tend to prefer ChatGPT, or are they more inclined to utilize Google? And what are the key factors influencing learners' preference to use ChatGPT for academic help-seeking? These relevant questions merit attention. The study employed a mixed-method research design to investigate university students' online academic help-seeking preferences. The results indicated that students tend to prefer using ChatGPT to seek academic assistance, reflecting the potential popularity of GenAI in the educational field. Additionally, in comparing seven machine learning algorithms, the Random Forest and LightGBM algorithms exhibited superior performance. These two algorithms were employed to evaluate the predictive capability of 18 potential factors. It was found that ChatGPT fluency, ChatGPT distortions, and age were the core factors influencing how university students seek academic help. Overall, this study underscores that educators should prioritize the cultivation of students' critical thinking skills, while technical personnel should enhance the fluency and reliability of ChatGPT and Google searches, and explore the integration of chat and search functions to achieve optimal balance.