Skip to content
My Code, My RulesMy Code, My Rules

ALL-WRIT0001 · ai_associated_pattern_count

Count exact AI-associated style patterns reported by external analyzers.

This is a deterministic rule for all languages. Read its implementation.

Definition

Section titled “Definition”

Read provider-neutral evidence generated by tools such as Vale AI Tells or Pangram. Count every selected pattern occurrence without combining phrase likelihood multiples. A provider baseline is retained as a separate measurement because observed count, expected count, and authorship probability are different quantities.

Evidence

Section titled “Evidence”

Each finding preserves the provider, provider version, external rule, matched text, source range, and any provider-reported relative likelihood. The configured provider filter is case-insensitive. An empty filter includes every provider. The value is the number of selected pattern occurrences across every eligible segment.

Exceptions

Section titled “Exceptions”

A pattern can be intentional, required terminology, quoted text, or ordinary prose in another domain or language. This rule reports selected house-style evidence and never claims that a person or model authored the text. providers selects which analyzers count, matched without regard to case, and an empty selection includes every provider the evidence holds. Context remains the deciding evidence.

Examples

Section titled “Examples”

A document with twenty Pangram evidence instances returns 20. If Pangram reports an expected baseline near four, the baseline stays a separate finding rather than being converted into an AI probability. A Vale alert for At its core contributes one count.

References

Section titled “References”