Abstract: Microbiological experiments are crucial in the teaching of life sciences and related majors, and microscopic sectioning and observation operations are particularly difficult and labor-intensive, involving various microorganisms such as bacteria, molds, actinomycetes, yeast, etc. Traditional experimental teaching often adopts a centralized teacher grading model, where students need to queue up and wait for manual review and feedback of microscopic images, resulting in high time costs and low efficiency. Teachers also spend a lot of energy on repetitive work. To address this bottleneck, this study developed the “Identify Microbe” model based on generative AI technology. Through customized training using prompt word strategies and workflow logic, the model can quickly recognize and automatically score microscopic images submitted by students within seconds, and output targeted operational improvement suggestions. After introducing AI models, the overall efficiency of the experimental classroom was significantly improved. Students could perform multiple rounds of smear operations and deeply contemplate the staining principle. The high score rate of microscopic image quality was significantly increased, and teachers were freed up more time for personalized tutoring and course expansion. The integration of the “Identify Microbe” model with microbiological experimental teaching not only resolved the pain points of batch evaluation of microscopic operations, but also provided practical and feasible ideas for the digital and intelligent upgrading of experimental education.

Key words: “Identify Microbe” model, artificial intelligence, microbiology experiment, experimental teaching, AI technology