Jev 在 Jozef Gherman 的 Jev Detector 為貼上的句子打 slop 分
分類- 延遲
- ~2 seconds for ~10,000 words
- 結果
- 只標示跨過證據門檻的句子
對每個句子問 Jev 0 到 4 的文體 Score，以及 Wikipedia 規準 Choice。
問題
Primitive
Choice
Ask a multiple choice question
- Name
- category
- 指示
- Which Wikipedia Signs of AI writing signal is strongest in this sentence? Pick the single best match.
- 準則
- none
Direct, specific writing without a notable pattern from the rubric.
- significance_puffery
Inflates importance, legacy, symbolism, or connection to a broader trend without adding concrete support.
- superficial_analysis
Adds a generic interpretive clause, often an -ing phrase, that merely restates why a fact matters.
- promotional_tone
Uses advertisement-like praise, grand adjectives, or unqualified claims of excellence and impact.
- vague_attribution
Assigns a claim to unnamed experts, observers, critics, reports, or studies without a checkable source.
- formulaic_language
Clusters stock AI vocabulary, negative parallelism, tidy triads, or ornate phrasing in place of specifics. One isolated word is not enough.
- canned_structure
Uses a generic conclusion, challenges-and-prospects ending, or outline-like transition that could fit many topics.
- reader_directed
Addresses the reader, describes the writing task, offers further help, or includes a chatbot-style disclaimer.
- formatting_artifact
Contains chatbot citation tokens, placeholder text, provider residue, or other direct generation artifacts.
- other_ai_style
Feels generic or model-like for a reason not cleanly covered by the other categories.
- none