AnalysisAI Models
2 days ago
Low diversity in LLM stories leads to repetitive 'Elias Thorne' tale
A study of 20,000 LLM-generated stories found 11 words appear in 88.3% of outputs, with minimal variation across models. This explains the widespread repetition of the lighthouse keeper 'Elias Thorne' story, highlighting low narrative diversity as a persistent limitation.