The development of generative AI is changing the user's alternative information search mode, prompting the user's preference to shift between the traditional human-mediated alternative information search mode and the new human-machine alternative search mode mediated by generative AI. Based on the problems and differences between the human-human alternative search mode and the human-machine alternative search mode, this paper provides an in-depth analysis of the key factors affecting the users' willingness to transfer between these two modes, which is of great significance for enhancing the efficiency of information search and optimizing the generative AI products.Based on the PPM (Push-Pull-Mooring) model, two models of users' willingness to transfer (willingness to transfer from human-human alternative search to human-machine alternative search and willingness to transfer from human-machine alternative search to human-human alternative search) were constructed, and the questionnaire survey was used to collect 407 and 430 valid sample data, respectively, and the subsequent data analysis and hypothesis testing were carried out using SPSS and AMOS software. The willingness to transfer human-human surrogate search to human-machine surrogate search is affected by a combination of push factors (dissatisfaction with information quality), pull factors (interactivity, anthropomorphism, accessibility), and anchoring factors (individual innovativeness, transfer costs).The willingness to transfer from human-machine to human-human surrogate search is influenced by a combination of push factors (dissatisfaction with service quality，privacy risk), pull factors (trust), and anchoring factors (social support).