Inclusive AI StrategiesAI Writing Flags Rubric
Scoring patterns common in AI-generated text, one flag per row
Prefer a document? Open the Google Doc version (opens in new tab). Google Docs will ask you to make your own copy, which you can then edit.
How to use this rubric
Use this rubric alongside your content criteria. Choose the rows that fit the assignment and set each row's value; the values below are suggestions. Score each row at its full value for Clear, half for Occasional, and zero for Pattern.
Some students were taught these patterns on purpose. Students who learned formula essays, transition-word lists or vocabulary drills in English language instruction often write "Moreover," "In conclusion" and "plays a crucial role." For those students, start with feedback and a revision before you deduct.
When a paper scores Pattern on three or more rows, hold a short conference before grading. Ask the student to explain one flagged sentence in their own words and tell you what they meant.
Build your rubric
Check Include for each row you want and set its points. The total updates as you go. Print my rubric prints only the rows you include, along with the notes on how to use the rubric and the student self-check. When your browser allows it, this page remembers your choices on this device.
Core flags
These rows follow the AI Writing Flags handout, flags 1 to 11.
| Include | Flag | Suggested value | Clear (full value) | Occasional (half value) | Pattern (no credit) |
|---|---|---|---|---|---|
| 1. Validation phrases ("That makes sense." "Of course you feel...") | 2 | Responds to the specific points in the prompt or text without reassurance openers. | One or two stock reassurances appear before the actual response. | Several paragraphs open with reassurance that answers nothing specific. | |
| 2. Reframing devices ("This isn't about X. It's about Y.") | 3 | States claims directly. Interpretations are labeled as interpretations ("I read this as..."). | One or two claims are made by rejecting a position, but each is also stated plainly. | The writer makes main claims by rejecting positions nobody in the source or prompt took. | |
| 3. False binaries ("One is a tool. One is a teacher.") | 3 | Gives a question as many sides as it has, including a continuum where one exists. | One or two tidy two-sided setups simplify the question. | Analysis is repeatedly forced into two sides, and details from the source that fit neither side are left out. | |
| 4. Dramatic setups ("Here's the thing:" "Honestly?") | 2 | Sentences and paragraphs open with substance. | One or two setups announce a point before making it. | Setups come before most key points. | |
| 5. Interrogation phrases ("Be honest." "Let me ask something gently.") | 1 | Questions, if any, are asked plainly. | One or two questions borrow an interviewer's or counselor's framing. | The writing repeatedly addresses the reader as if interviewing them. | |
| 6. Authority markers ("I'll be honest." "Real talk.") | 1 | Sentences are direct without announcing it. | One or two announcements of honesty or plain speaking. | Announcements of honesty or plain speaking appear throughout. | |
| 7. Callback devices ("You said it yourself.") | 2 | Earlier points and sources are referred to by their content. | One or two callbacks point back without naming what was said. | Repeated callbacks suggest depth without connecting to a specific point. | |
| 8. Aphoristic flourishes ("Your data travels with you.") | 3 | Literal verbs; headings and topic sentences describe what follows. | One or two tagline phrases or objects given motion or intent. | Key points are written as slogans; the plain claim is hard to find. | |
| 9. Negation cascades, narrative writing only ("Not anger. Not sadness. Just...") | 2 | Feelings are shown through an action, object, or sensation the reader can picture. | One cascade, or one that ends on an abstraction. | Repeated cascades stand in for description. Mark this row N/A for non-narrative writing. | |
| 10. Promissory clauses ("...and that matters more than it seems.") | 2 | Points are delivered without announcing that they are coming or rating them. | One or two clauses promise or rate a point instead of making it. | Promises and self-ratings appear in most paragraphs. | |
| 11. Em dash overuse | Feedback only | Punctuation states the relationship: comma, colon, parentheses, or a new sentence. | Several long dashes in a single paragraph. | Long dashes join most sentences. Many human writers do this; give feedback rather than deduct unless the class has practiced the alternative. |
Extended flags
Optional rows, drawn from the same pattern research. Add the ones that fit the assignment.
| Include | Flag | Suggested value | Clear (full value) | Occasional (half value) | Pattern (no credit) |
|---|---|---|---|---|---|
| 12. Meta-commentary ("It's important to note that...") | 1 | Points are made without commenting on them. | One or two phrases announce that a point matters. | Most points are introduced by commentary about the point. | |
| 13. Stock transitions ("Moreover," "Furthermore," "In conclusion") | 1 | Sentences connect by meaning, with plain transitions ("and," "but," "so") where needed. | Formal transitions open a few paragraphs. | A formal transition opens nearly every paragraph or sentence. | |
| 14. Inflated vocabulary ("delve," "tapestry," "foster," "leverage," "underscores") | 1 | Plain words name specific things. | A few inflated words where a plain one would do. | Inflated words cluster in most paragraphs. If the student was taught these words, start with feedback. | |
| 15. Generic intensifiers ("plays a crucial role," "highly effective") | 1 | Importance is shown with evidence, a result, or a specific role. | One or two intensifiers stand in for evidence. | Intensifiers carry most claims of importance. | |
| 16. Time fillers ("In today's world," "Throughout history") | 1 | Paragraphs open with the actual point. | One opener of this kind. | Several paragraphs open with a filler about time or society. | |
| 17. Triplet cadence (lists of three, the last item as the punchline) | 1 | Lists have as many items as the content has. | A few lists of three where the third item is decoration. | Most paragraphs are built around a list of three. | |
| 18. Uniform structure | 1 | Paragraph length and shape follow the content. | Some paragraphs repeat the same shape (claim, three supports, uplifting close). | Every paragraph has the same length and shape, regardless of content. | |
| 19. Closing uplift | 1 | The ending states a conclusion the paper supports. | The ending adds a general uplifting line after the conclusion. | The ending is a general inspirational statement that could close any paper. | |
| 20. Generic examples | 3 | Includes at least one detail only this student, in this class, could have written: a personal experience, a class discussion, a local place, a specific source passage. | Examples are accurate but could belong to any student. | No example ties the writing to the student, the course, or the community. |
Cluster check
Three or more flags in one paragraph usually means the paragraph needs rewriting from its main point. Highlight the cluster, name the point you think the student meant, and ask the student to write that sentence first.
Student self-check
Run these on your draft before you submit:
- Delete test: remove a phrase you are unsure of and reread. If nothing is lost, leave it out.
- Tagline test: if a sentence would work on a poster, rewrite it as information.
- Picture test: after a string of "not this, not that," can you picture what the sentence finally names? If not, find the image.
- Own-detail test: point to one detail in your paper that only you, in this class, could have written.
Rubric from Inclusive AI Strategies: inclusiveaistrategies.com/ai-writing-flags-rubric
Related
Questions?
For questions, email info@inclusiveaistrategies.com
Interested in an in-person or virtual presentation/workshop? View some presentation topics