This study leverages the technical methodologies of comparative analysis and quantitative assessment to scrutinize the role of Artificial Intelligence in English writing evaluation, with a particular focus on Artificial Intelligence Generated Content (AIGC) assistance. The research object comprises 110 English compositions from CET-4 mock exams of freshmen and sophomores with the evaluations of five seasoned teachers serving as a control group. Employing the ChatGPT large language model for AIGC evaluation, we meticulously designed specific prompts and scoring criteria aligned with national CET-4 composition assessment standards. The research process entailed a multifaceted quantitative assessment of evaluation quantity, dissecting evaluation information into discrete units and applying the Mann-Whitney U test to ascertain median differences. Evaluation types were analyzed through the construction of a comprehensive coding table, conversion of raw counts into median proportions, and subsequent Mann-Whitney U test for comparative analysis. Evaluation accuracy was rigorously compared by calculating precision and recall rates across four critical dimensions: spelling, punctuation, word usage, and grammar. To gauge the qualitative perspective, semi-structured interviews were conducted to gather teachers’ insights on the quality of ChatGPT-generated evaluations. The findings reveal that AIGC, while offering a higher volume of evaluations and diverse evaluation types, exhibits lower precision rates but higher recall rates in certain aspects. Educators acknowledge the advantages and limitations of AIGC, suggesting its potential to assist in composition evaluation, albeit with the necessity for algorithmic refinement. This study provides educators with a reference framework for the enhanced application of AIGC in writing instruction, emphasizing the imperative for algorithmic enhancement to harness the full potential of AIGC in educational assessment.