This study aims to conduct a comparative evaluation of the efficiency of three editing approaches in Persian texts: human editing, automatic editing relying on existing proofing tools, and intelligent editing based on generative artificial intelligence. Methodologically, the research is qualitative in nature and adopts an exploratory approach. In this regard, the identified errors and the corrections made by all three agents were examined across technical, morphological, syntactic, semantic, and content-related domains, and the strengths and weaknesses of each were identified. The evaluation criteria included accuracy in error detection, correctness of the suggested corrections, fluency of the proposed text, and adherence to content structure. Three texts from different domains (practical, Islamic sciences, and modern sciences) were selected as sample texts to assess the performance of these three methods across various linguistic and thematic contexts. The findings indicate that automatic editing performs relatively accurately in detecting formal errors and spacing issues but is ineffective in addressing morphological, syntactic, and semantic errors as well as correcting incorrect expressions. Intelligent editing proves more successful in text fluency improvement, sentence shortening, punctuation, and paragraphing; however, it makes errors in understanding specialized contexts and accurately identifying technical terms. Provided that human editing is carried out properly, it constitutes the most accurate and reliable model for correcting complex errors, understanding the semantic layers of the text, and adhering to the orthographic and stylistic norms of the Persian language. However, given the challenges associated with human editing, it is recommended to employ automatic editing and, in particular, intelligent editing. The norms of Standard Persian can be supplied to generative artificial intelligence systems, and the capabilities of intelligent editing can be utilized under the supervision of a specialized editor.