How to Identify AI-Assisted Writing Patterns

A practical guide for teachers reviewing a student’s writing process report

This guide lists common patterns to help you understand how a student completed a writing task and whether parts of the work involved AI assistance. These indicators should be used as conversation starters, not conclusions.

1. Look at editing time

Editing Time shows how long the student was actively typing or revising, not counting breaks. Very short editing time (for example, completing a 500-word essay in two or three minutes) is unusual and worth reviewing.

Use the Editing Time section of the report to see:

• Total active writing time
• Whether the duration matches the length and complexity of the assignment

Short editing time does not prove AI use, but it can signal that text may have been pasted or prepared elsewhere.

The Editing Time section in a writing process report shows editing time in the document.

2. Review large pasting events

Students sometimes paste AI-generated text into their document. The Paste Events section lists all paste events and how many characters were added.

What to look for:

• Sort by the Characters column to bring the largest pastes to the top
• Check whether the Source column says “Likely from Outside”
• Click on Pasted Text to see exactly what was inserted
• Click the Time of Paste to jump to the comparison view and see where in the document the pasted text appears

Large blocks of text added instantly are often worth inspection or discussion.

3. Look at text deletions

The Text Added and Removed chart shows the rhythm of writing. It shows what was typed and what was deleted.

A healthy writing process usually includes:

• Small bursts of typing
• Occasional pauses
• Frequent small deletions or revisions

Possible signals worth exploring:

• A chart that is almost entirely green (text added) with no red (text removed)
• Many green bubbles of identical size appearing in a row (but this could also indicate focused free writing)
• A few very large green bubbles where huge amounts of text appear at once

You can click any bubble to see exactly what changed.

4. Inspect breaks taken

The Editing Sessions and Breaks section shows when the student typed and when they paused.

This can give insight into writing habits, such as:

• Working steadily over time
• Completing everything at the last minute
• Taking long breaks followed by quick bursts of activity

This information does not directly show AI use, but last-minute writing patterns sometimes correlate with increased reliance on AI tools.

5. Inspect free typing and revision patterns

The Edit Time and Location chart visually combines three things:

1. When the student made an edit
2. How much text changed
3. Where in the document the edit occurred

This chart gives a fast read on the writing journey once you use its options:

• Switch granularity to “words” (instead of the default “sentences”) to surface tiny edits that may be hidden.
• Turn on coloring by sentences or paragraphs to see which passages were revisited and revised.

Here are some common patterns that this chart could reveal:

• A smooth diagonal line formed by many small dark bars. A smooth diagonal line from the top right corner to the bottom left corner formed by many small dark (or colored) bars indicates continuous typing from start to finish. It can look like a projectile or a waterfall from left to right. This could indicate focused free writing or retyping another source.

• One or more tall dark bars. One or more tall dark (or colored) bars indicate large amounts of text added around that time.

• Short dark bars scattered in a single time point. Short dark (or colored) bars clustered at a single time point suggest many small changes made in different places around the same time. This could happen from accepting spelling/grammar suggestions. This pattern is visible only at “word” or “character” granularity, not in “paragraph” or “sentence” granularity.

• Sudden decrease in full bar height. A sudden decrease in the overall bar height, compared to the bar immediately to the right, indicates text deletions.

• Dark bars scattered across the chart. Dark (or colored) bars scattered all over the chart suggest lots of revision throughout the document. For example, a mix of continuous typing at first followed by scattered edits later can indicate drafting followed by revision (something typical of genuine writing).

6. Increase the detail when needed

By default, Process Feedback summarizes the writing process using up to 360 timepoints. This spacing usually provides enough detail for reports with editing time under 15–20 minutes. This default setting ensures that the report loads in a reasonable time in most devices.

For document with longer writing history, you may want a more detailed view with higher number of timepoints. On the report’s opening screen, choose a higher option in the “Timepoints” dropdown. This increases detail and shows smaller edits more clearly.

Uncommon Scenarios

1. Retyping an AI-generated text

Some students may generate an essay using AI and then retype it manually so their process looks “authentic.” This often appears as:

• A perfect, uninterrupted diagonal line formed by dark (or colored) bars in the Edit Time and Location chart
• Very few deletions shown in the Text Added and Removed chart
• Consistent, uninterrupted typing rhythm with approximately consistent typing speed in the Typing Speed Timeline chart

This pattern does not guarantee misconduct, but it may be uncommon for original writing.

2. Voice typing

Some students use voice typing:

• As an accessibility accommodation
• Because it feels easier than typing

Process Feedback identifies voice-typing events automatically when used in Google Docs. Voice typing can also look like paste events because text appears in chunks.

3. AI agents mimicking a student

Some AI tools can generate text that imitates an individual student’s style. These cases are hard to detect from a process report alone.

The Most Reliable Check

The most dependable way to confirm authorship is to meet briefly with the student and ask them to produce a similar short piece of writing in person. You can then compare their natural writing process with the submitted report.

Once again, these indicators should be used as conversation starters, not conclusions.

Our Perspective

This guide is created to help dozens of high-school teachers who requested us to create it. At Process Feedback, we neither endorse nor oppose students using AI—our technology simply reveals the steps a student took to complete their work. You decide how to use these insights. We believe that a student’s working process matters, regardless of whether AI is involved. Feedback on the process should be context-specific, which is why we don’t provide explicit evaluations. Instead, we make it easier to offer or acquire meaningful process feedback.

Related Resources

1. Dr. Colin Allen’s blog article “Teaching Writing in the Age of LLMs”. Dr. Allen is a distinguished professor in the Department of Philosophy at the University of California, Santa Barbara.