In the age of ChatGPT, “showing your work” has taken on new meaning and importance.
As teachers and professors look for ways to use AI to prevent homework cheating, many are starting to ask students to share the history of their online documents to check for signs of bot writing. . In some cases, this may include asking students to allow access to the document’s version history in systems such as Google Docs, or leveraging new web browser extensions created for this purpose. Masu.
Many educators use this approach, often referred to as “process tracing,” instead of running student work through an AI detector. AI detectors tend to falsely accuse students, especially those whose first language is not English. Even companies selling AI detection software admit that their tools can misidentify student-written material as AI. with a probability of about 4%. Teachers grade so many papers and assignments that many educators consider this an unacceptable level of error. And some students have faced backlash with social media posts. school sued About what they claim is a false accusation of AI fraud.
The idea is that a quick look at the version history can reveal whether a large amount of posts were suddenly pasted from ChatGPT or other chatbots, and that this method is more reliable than using AI detectors. That’s it.
But as process tracking becomes more popular, a growing number of writing teachers are speaking out against it, arguing that the practice amounts to surveillance and violates students’ privacy.
“It casts doubt on everything,” said Leonardo Flores, a professor and chair of the English department at Appalachian State University in North Carolina. He was one of several professors to outline arguments against the practice. blog post Last month, a joint task force on AI and writing was organized by two prominent academic organizations: the Modern Language Association and the Conference on University Composition and Communication.
Could process tracking be the answer to verifying the authenticity of student work?
Time lapse history
Anna Mills, an English instructor at the College of Marin in Oakland, California, uses process tracking in her writing classes.
For some assignments, she asked students to install and allow access to a web browser extension called Revision History. With this tool, a ribbon of information appears above the student’s submitted document showing time spent and other details of the writing process. The tool also generates a time-lapse video of all the typing entered into the document for teachers to view, giving them a rich behind-the-scenes look at how the essay was written.
Mills also gave students access to a similar browser plug-in feature called Authorship that Grammarly released in October. Students use the tool to Generate a report Information about the creation of a particular document, including details about the number of times the author has pasted material from another website and whether the pasted material may have been generated by AI. You can also create time-lapse videos while creating documents.
Instructors tell students that they can opt out of tracking if they have concerns about this approach, and that they will find another way to verify the authenticity of their work. But no student has brought it up with her yet, and she wonders if they’re worried that asking them to do so will make them look suspicious.
Most of her students seem receptive to tracking, she says. In fact, some students have called for even stronger checks on AI fraud in the past. “Students know that there is a lot of AI fraud going on, and that they risk having their jobs and degrees diminished as a result,” she says. And while she believes the majority of students create their work themselves, she says she has witnessed students submit AI-generated work as their own. “I think some accountability makes sense,” she says.
But other educators argue that having students see a complete history of their work increases self-awareness. “If I were a student and I wanted to be sure that my process was being tracked and that that information was somehow within the professor’s purview, I would have to share my process, or worse, In particular, you’ll feel self-conscious and anxious that your process is being judged — my writing,” said Kofi Adisa, associate professor of English at Howard Community College in Maryland. wrote in a blog post of the Academic Committee on AI.
Of course, students could very well use these AI tools at work and move into a world where they have to show employers what parts of their work they create. However, Adisa said, “With more and more students using AI tools, I wonder if some instructors are relying too much on writing monitoring rather than actual writing instruction.”
Another concern raised about process tracking is the possibility that some students may do things that look suspicious but benign to process tracking tools, such as drafting a section of a paper and pasting it into a Google Doc. There is.
For Appalachian State’s Flores, the best way to combat AI plagiarism is to change the way teachers design assignments so that they accept the fact that AI is now a tool that students can use, rather than something that is forbidden. . Otherwise, he says, there will simply be an “arms race” of new tools to detect AI and new ways students devise ways to circumvent those detection methods.
In theory, Mills doesn’t necessarily disagree with that argument. She discusses what experts are advising teachers to do – overhaul the way they teach – and the reality of how educators are rushing to adopt AI to eradicate rampant cheating. He says he feels there is a big gap between the two approaches.
“We’re at a point right now where there are a lot of potential compromises and a lot of competing forces that teachers don’t have much control over,” Mills said. “The biggest factor is that the other things we recommend require a lot of institutional support, professional development, effort and time,” which most educators don’t have.
product arms race
Grammarry officials say they’re seeing high demand for process tracking.
“This is one of the fastest growing features in Grammarly’s history,” says Jenny Maxwell, the company’s director of education. She says customers have generated more than 2 million reports using the process tracking tool since it was released about two months ago.
Maxwell said the tool was inspired by the story of a college student who used Grammarly’s spell check feature on his paper and was falsely accused by his professor of using an AI bot to write his paper. The student, who said he lost his scholarship due to cheating accusations, shared details of the incident in a series of TikTok videos that went viral and eventually led to him becoming a paid consultant for the company.
“Marley is like our North Star,” Maxwell says. The idea behind Authorship is that students can use the tools while writing, so if they’re accused of improperly using AI (in Marley’s words), they can report it as a way to denounce the professor. This means that you can submit. “It’s really like insurance,” Maxwell says. “If it gets flagged by AI detection software, you actually have proof of what you did.”
Regarding student privacy, Maxwell emphasized that the tool is designed to give students control over whether they use this feature, and that students will be able to view their reports before handing them over to their instructors. This is in contrast to a model where professors run student papers through an AI detector. Students rarely see reports that say which sections of their work were written by AI.
Turnitin, one of the most popular AI detectors, is also considering adding process tracking capabilities, said Annie Chechitelli, Turnitin’s chief product officer.
“We’re looking at what are the elements that make sense for students to demonstrate that they did this on their own,” she says. The best solution may be a combination of AI detection software and process tracking, she added.
She argues that leaving it up to students to enable process tracking tools may not do much to protect academic integrity. “Opting in makes no sense in this situation,” she argues. “If I’m a fraud, why would I use this?”
Meanwhile, other companies are already selling tools they claim can help students overcome both AI detectors and process trackers.
Mills, of the College of Marin, said she recently heard about a new tool that allows students to paste AI-generated papers into a system that simulates students typing their papers character-by-character into process-tracking tools like Authorship. Ta. Changing the keystrokes makes it look more authentic.
Chechitelli said the company is closely monitoring the proliferation of tools that “humanize” AI-generated writing and allow students to submit their work as their own without detection.
She says she’s surprised by the number of students posting TikTok videos boasting that they’ve found a way to destroy AI detectors.
“It helps us. Are you kidding? That’s great,” Checchitelli says. He believes social media posts like this are the easiest way to learn techniques and modify products accordingly. “You can see which ones are getting traction.”
