2026-05-29 08:22 UTCSummary onlyUpdated: 2026-06-30 13:03 UTC

StoryScope: Investigating Idiosyncrasies in AI Fiction

A new study introduces StoryScope, a method that distinguishes AI-generated from human-written stories by analyzing narrative structure rather than writing style. Using a corpus of 61,608 stories with 304 features each, the approach achieves 93.2% macro-F1 for human vs. AI detection and reveals distinct narrative fingerprints for different LLMs like Claude, GPT, and Gemini.

SourceHacker News AIAuthor: ironyman

Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions

Summary only

Under the source content permission, this page provides an editorial summary, key points, and a link to the original article.

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

StoryScope extracts discourse-level narrative features (e.g., character agency, temporal discontinuity) to differentiate AI fiction from human writing, without relying on stylistic cues.
On 61,608 stories (~5,000 words each), narrative features alone achieve 93.2% macro-F1 for human vs. AI detection and 68.4% for six-way authorship attribution.
AI stories tend to over-explain themes and favor tidy, single-track plots, while human stories feature more morally ambiguous choices and greater temporal complexity.
Different LLMs show unique narrative fingerprints: Claude has flat event escalation, GPT overuses dream sequences, and Gemini defaults to external character descriptions.

Highlights and analysis are generated automatically and may contain errors. Check the original source.

Summary only

Article intelligence

Key points

More recent stories on this topic · Models