Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI news ·

Georgia Tech tool shows how AI shapes student writing in real time

Georgia Tech and Stanford researchers built DraftMarks, an open-source tool that visually maps how students use AI while writing. It shows where AI text appeared, what was kept, and what the writer changed.

Georgia Tech tool reveals how AI shapes student writing

A new open-source tool called DraftMarks makes visible what's typically hidden: how students use AI while writing. Developed by researchers at Georgia Tech and Stanford, it marks up documents to show where AI was involved, how heavily, and what the writer changed afterward.

The tool addresses a real problem. Nearly 90% of college students now use AI in coursework, with almost half using it during drafting. Traditional detection tools like Turnitin can flag AI's presence, but they don't show whether a student engaged critically with it or just accepted whatever the tool generated.

How it works

DraftMarks uses visual cues layered directly onto text. Eraser crumbs mark heavily revised passages. Masking tape highlights AI-generated sections. Glue residue shows where AI text was removed. Ghost text appears when a writer prompted AI but rejected the output. Different fonts distinguish human writing from machine-generated content.

The marks tell a story about the writing process itself. They show not just that AI was used, but how-and whether the writer made intentional choices about it.

Behind the scenes, the tool tracks a document's draft history and classifies different types of edits and AI interactions in near-real time.

What educators actually want to see

The researchers didn't start by building detection algorithms. They started by watching how instructors actually review student work. In a 21-person study, they observed what cues teachers looked for when assessing learning, revision, and originality.

Those insights shaped DraftMarks' visual language. The marks deliberately mimic physical artifacts of writing-tape, eraser debris, smudges-because instructors already recognize these signals from paper drafts.

In a follow-up study with 70 participants, instructors were most interested in seeing how ideas developed and where students exercised judgment. General readers used the marks differently: to assess trust and authorial intent, deciding how much confidence to place in a piece of writing.

Reflection instead of detection

Unlike AI detectors that spit out a percentage, DraftMarks prompts writers to think about their choices. One researcher noted that seeing AI's role in his own writing made him reconsider his tone and realize how small AI choices can subtly reshape what he was trying to say.

The tool shifts the conversation from "Did the student use AI?" to "How did the student use AI, and did they do it intentionally?"

For writers working with AI, understanding your own process matters. Learn more about AI for Writers and how to use these tools effectively. For educators, AI for Education covers assessment strategies in the age of generative AI.

Share