Stupid AI fiction stories share these common quirks, study says

A study identified quantifiable stylistic patterns that distinguish AI-generated fiction from human-written stories, including overuse of words like 'tapestry' and an excess of dream sequences. Researchers found these quirks are consistent enough across AI outputs to be measurable and potentially detectable. The article, published on IBM Think, explores whether readers can identify AI-generated prose based on these recurring idiosyncrasies. The findings have implications for content authenticity, LLM output quality, and the broader question of what 'good' generative AI writing looks like.