Key points are not available for this paper at this time.
This study examined the readability and stylometric characteristics of AI-generated texts (AIGTs) produced by ChatGPT and DeepSeek. Specifically, it examines the extent to which AIGTs differ in terms of readability, morphological structures, lexical choices, syntactic patterns, and semantic features. Using Open Brain AI, the study analyses 10 essays generated by the two models, forming a corpus of ∼10,000 words. Readability analysis showed that ChatGPT uses longer sentences than DeepSeek. While DeepSeek produces texts suitable for high school students, ChatGPT’s texts are more appropriate for college students. Stylometric analysis indicates that ChatGPT demonstrates a higher density in adjectives, adpositions, auxiliary verbs, subordinating conjunctions, and verbs, while DeepSeek shows a higher density in adverbs, coordinating conjunctions, nouns, particles, and pronouns. In addition, ChatGPT-generated texts exhibit richer and more diverse lexicons. Additionally, syntactic analysis showed that ChatGPT leads to the use of adverbial modifiers and direct objects. By contrast, DeepSeek leads determiners, nominal subjects, objects of prepositions, and roots. Finally, the findings reveal that DeepSeek outperforms ChatGPT in incorporating semantic features, except when events and ordinal and cardinal numbers are used. These insights highlight the distinctive characteristics of each AI tool, which can inform strategies for enhancing writing style.
Jaashan et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: