A rhetorical device commonly found in AI-generated text has been identified as a notable feature of writing, particularly in the context of famous literary works. This construction, which contrasts two ideas—often phrased as "It's not X; it's Y"—has been observed in various contexts, including corporate communications and social media posts. For instance, a report by Citizens Financial Group noted that this structure has become increasingly prevalent, with its usage in corporate communications reportedly quadrupling from 2023 to 2025. Researchers at Pangram estimate that such sentences appear three times more frequently in AI writing compared to human writing.
The article discusses the implications of this pattern, noting that while it can be effective when used sparingly, it has become formulaic in AI outputs. OpenAI's product manager, Laurentia Romaniuk, referred to this construction as "contrastive phrasing" and acknowledged that the company is working on diversifying the chatbot's language use.
The reasons behind the prevalence of this rhetorical device in AI writing remain unclear. Some experts suggest that it may stem from the patterns identified in the vast amounts of human-written text used to train AI models. Additionally, the reinforcement learning process, where human reviewers grade AI responses, may inadvertently favor this style of writing.
The article also highlights the potential for a cycle of reinforcement, where AI-generated text continues to influence future models, potentially leading to a homogenization of writing styles. Despite the challenges posed by these patterns, they may also serve as markers that distinguish AI writing from human writing, as the distinctiveness of such constructions persists in AI outputs.