With the increasingly widespread application of artificial intelligence (AI) technologies, more and more people are concerned about whether machine intelligence will replace traditional human labor. However, while there are concerns, the relationship between AI and traditional labor is not entirely mutually exclusive. They can cooperate under certain conditions. The BP neural network was applied to conduct a comprehensive evaluation of ChatGPT performance satisfaction, analyze the public's satisfaction with ChatGPT's performance, and provide references and suggestions for the future co-development of ChatGPT with humans. A satisfaction survey was conducted among 11 high-tech industrial parks and university students in Hangzhou, exploring the public's satisfaction with ChatGPT's generated language quality, knowledge accuracy, context understanding ability, model response speed, controllability and interpretability. In the prediction of public satisfaction, the BP neural network performs well in four indicators, such as accuracy, recall, harmonic mean and F1-score during model testing. The result shows that ChatGPT's context understanding ability, controllability and interpretability are important influencing factors of ChatGPT performance satisfaction. In the future, ChatGPT should further strengthen training and optimization in these three aspects.