Exploration and analysis of generative artificial intelligence in editing and proofreading:Based on practical tests of ChatGPT and over 150 Chinese large models
[Purposes]This study systematically reviews and tests the performance of currently available large models on the internet in terms of editing and proofreading.It aims to clarify the strengths and weaknesses of the existing large models,provide references for editors choosing large models for editing and proofreading,and offer a basis for advancing the development of these models'capabilities of editing and proofreading.[Methods]Different types of texts with varying complexity levels of error were designed to evaluate the accuracy and stability of the models'responses.Methods such as text comparison,comparative analysis,and statistical analysis were comprehensively employed.[Findings]58 models demonstrate editing and proofreading capabilities.The study showcases the performance of 36 models when handling different types and levels of textual complexity,summarizes the shortcomings observed during testing,and shows that the Chinese models have comparative advantages over ChatGPT.[Conclusions]In practical aspects,editors can select appropriate models to assist with their tasks,establish knowledge bases and personalized model-based editing and proofreading methods,use role setting and chain of thought inquiry methods to improve efficiency,and further enhance their information literacy and professional skills.
Large modelArtificial intelligenceEditing and proofreadingPractical test
NETL
NSTL
万方数据
