Advancements in generative AI has made it harder for teachers to know how a student arrived at a finished essay. A paper may be grammatically polished, well organized, and seemingly complete, but did the student arrive there authentically? WriteUp offers a way for teachers to see the process behind the writing.
Get Started
The ideal writing process is non-linear. We begin without knowing exactly where we are going. We write something, reconsider it, change direction, delete parts, return to earlier ideas, and discover something we didn't initially know we thought. The process is a simularcrum of the author's thinking. By continually reworking the same piece of writing is how we arrrive at a better version of our own thinking.
Writing is therefore a process of discovery. This is the benchmark for writers at every level.
There are no shortcuts to it. In education, going through this process again and again is part of learning to think. Students aren't simply producing essays. They are learning how to develop an idea, interrogate their own thinking, make decisions, and communicate something that is genuinely their own. The finished essay is only the endpoint of that process.
Teachers, however, are often only given the endpoint. They receive a finished essay and have to determine whether the student actually went through an authentic writing process to produce it.
Generative AI makes the problem substantially more complicated. A student can now enter a prompt and receive a polished version of an essay almost immediately. The final document may look like the product of a thoughtful writing process without much input from the student. This creates a nearly impossible task for teachers. They are asked to evaluate not only what a student wrote, but whether the student actually did the thinking required to write it.
A student who carefully develops an essay over several hours and a student who generates an essay in seconds can potentially submit documents that look remarkably similar.
This also creates a problem for students who genuinely wrote their own work.
Content-based AI detection can produce false positives, meaning students who went through an authentic writing process can find themselves having to prove that their work is their own.
Students using AI to write papers can also cover up their own tracks by humanizing their writing, making it undetecatable by these content-based AI detection tools.
Instead of asking only "Does this essay look like AI?", we can ask a more fundamental question:"Does this look like someone actually writing?"
Authentic writing leaves traces of the thinking that produced it. Students move between sections. They write and delete. They revise sentences. They change their minds. They return to earlier ideas. They experiment with language. They sometimes spend several minutes working through one sentence and then suddenly write several paragraphs.
The process is messy because thinking is messy. This non-linearity is not a flaw in writing. It is one of the defining characteristics of authentic writing. By observing the process rather than only the finished product, teachers can gain information that a final essay cannot provide.
WriteUp measures patterns in the writing process through a series of metrics designed to capture this non-linearity.
Rather than attempting to determine whether a particular sentence sounds like AI, these metrics examine how the student actually wrote the document.They can capture patterns such as:
How a student moves back and forth through a document; How frequently they revise and delete; How their writing develops over time; The relationship between writing and revision; Patterns in how a student builds and reworks ideas.
Together, these patterns create a Process Score: a way of understanding how much evidence there is of an authentic writing process behind the final document.This provides a fundamentally different signal from content-based AI detection.
The goal isn't to replace a teacher's judgment with another algorithm. It is to give teachers better information.Teachers already know their students in ways that no AI detector can fully capture. They hear how students talk in class. They see how they develop ideas. They read their previous work. They understand their strengths, weaknesses, interests, and ways of expressing themselves. The writing process gives teachers another piece of that picture.
Instead of spending their time worrying about whether an essay was authentically written, teachers can cross-reference a student's Process Score with their own understanding of that student.
A student whose writing process shows sustained development, revision, and non-linear thinking provides teachers with evidence that the final essay came from an authentic process.
The goal shouldn't be to build a better system for catching students. It should be to make writing process visible. Allowing for teachers to get a better understanding of their students writing behaviors. Think of it like an X-ray into their student's writing.
We are constantly building new ways to help remove the AI detection burden from teachers and give new teaching tools to teachers and students.
