Drivers and Risks of Generative AI Technology in Research Innovation: A Survey on Technology Acceptance and Usage Behavior of Researchers

Abstract

Since 2018, global AI technology has entered a new phase of explosive growth, with continued breakthroughs in language, vision, and multimodal macromodels, among many other areas. 2025 Generative AI, represented by the DeepSeek model, is emerging as a powerful reinforcement learning capability that provides researchers with a highly flexible space for adapting technologies and promotes interdisciplinary collaboration. The AI technology has been a major factor in accelerating scientific research. However, while AI technology accelerates the scientific research process, it also triggers a series of contradictions. On the technical level, the black box of algorithms and the lack of model interpretability have called into question the reliability of scientific research conclusions; on the ethical level, the problem of data privacy leakage and the controversy over the right of authorship of academic results are growing. Based on the above background, this study focuses on researchers in universities in Henan Province, and comprehensively utilizes various methods, such as questionnaire survey, support vector machine, and structural equation modeling, to explore the current status of generative AI application in academic research. The results of the study show that cognitive differences, researchers in the field of natural sciences and engineering technology due to high frequency used for experimental design, data processing and recognized efficiency; humanities field use and satisfaction is lower. In terms of satisfaction, the literature query function is highly satisfactory, the research planning and design and data processing and analysis functions are moderate, and the academic article writing function is low. The findings of this paper provide theoretical support and practical insights for the popularization and application of AI technology.