Since the sudden rise of ChatGPT and other AI chatbots, many teachers and professors have started using AI detectors to check their students' work. The idea is that a detector will detect if a student has the robot do a job.
However, this approach is controversial as these AI detectors have been shown to return false positives. In some cases, a student may claim that the text was generated by her AI even though she did all the work herself without the assistance of a chatbot. False positives appear to occur more frequently in students who do not speak English as their first language.
As a result, some instructors are trying different approaches to prevent AI cheating. It is borrowed from a criminal investigation example.
This is called “linguistic fingerprinting,” and it uses linguistic techniques to determine whether a text was written by a particular person based on an analysis of previous writing. The technology, also known as “author identification,” was discovered when an analysis of Kaczynski's 35,000-word anti-technology manifesto was cross-checked with his previous writings, including Ted, the terrorist known as the Unabomber for a series of deadly mail bombs. – Helped arrest Kaczynski. Help identify him.
Mike Kentz was an early adopter of the idea of bringing this fingerprinting technology into classrooms, claiming that this approach “flipped the script” on the usual methods of checking plagiarism and AI. He is an English teacher at Benedictine Military Academy in Savannah, Georgia, and also writes a newsletter about the challenges AI poses to education.
Kentz shares his experience with this approach and discusses the pros and cons on this week's EdSurge podcast.
Listen to this week's episode in full. Listen on Apple Podcasts, Overcast, Spotify, or wherever you listen to podcasts. Alternatively, use the player on this page. Or read the partial transcript below, lightly edited for clarity.
EdSurge: What is linguistic fingerprinting?
Mike Kents: This is very similar to a normal fingerprint, but it is related to the way it is written. And that's the idea that each of us has a unique way of communicating that can be patterned, tracked, and identified. If you have a known document written by someone, you can pattern the fingerprints written by that person.
How is it used in educational settings?
If you have a document that you know was written by a student, check the new essay submitted by the student against the original fingerprint to see if the language style matches the syntax, word choice, and vocabulary density. can. …
There are also tools to create reports. And it's not saying, “Yes, this student wrote this,” or “No, that student didn't.” It's on a spectrum, and there are a bunch of vectors in the system that are on a kind of pendulum. Find out the percentage chance that the author of the first paper also wrote her second paper.
It seems that there was a time recently when this approach was useful at your school. Could you please share that?
My freshman science teacher came to me and said, “Hey, I have a student who wrote something that doesn't look like him.'' Have you written anything else? That way I can be sure that I don't blame him when he doesn't deserve it. ” And I said, “Yes, sure.”
and we did it [linguistic fingerprint tool] And it produced a report. The report confirmed what the student believed was not written.
The biology teacher went to the mother and said that she didn't even have to use the report, but it didn't look like the student had written the report. And it turned out that his mother wrote it for him, more or less. In this case, it wasn't AI, but the truth is he just didn't write it.
Some critics of this idea point out that students' writing should change as they learn, so fingerprints based on previous writing samples may no longer be accurate. Shouldn't students' writing be changed as well?
If I've ever taught middle school writing, or if I've taught early high school writing, their writing doesn't change much in eight months. Yes, I hope it improves. Yes, it will get better. But what we're talking about is a very sophisticated algorithm, so even if you have some great writing teachers, not much will change in 8 months. You can also do new assignments at any time to get a fresh “known document” of their writing later in the semester.
Since this technique comes from law enforcement, some may worry that it has some sort of criminal justice vibe.
If a situation arises next year where I think my child may have used AI, I would not immediately go through the fingerprinting process. That's not the first thing I do. Let's talk to them first. There's enough trust there and I hope we can work something out. However, I think this is just a good backup just in case.
Schools should have a system of rewards and consequences, a system to enforce rules and discipline children when they deviate from discipline. for example, [many schools] There's a camera in the hallway. So you're doing this to document evidence in case something goes wrong. We have a full range of disciplinary measures in place backed by mechanisms to ensure that they are actually upheld.
How optimistic are you that this or other approaches you're trying will work?
I think it's going to be a very turbulent situation for the next five years or so, maybe even longer. I think the Ministry of Education and local governments need to establish AI literacy as a core competency in schools.
And we need to change our assessment strategies, change our children's interest in creative writing, and realize that the written work is actually no longer useful. You know my new thing is also oral communication. So when my child finishes writing an essay, I now say, “Okay,'' and write more. Everyone just goes up to the venue without any documents and speaks for his 3-5 minutes about his case, or whatever it is. Your job is to verbally communicate what you were trying to claim and how you proved it. Because that's something AI can't do. So my optimism lies in rethinking valuation strategies.
What I fear more is that trust in the classroom will be destroyed.
I think schools will face big problems next year. There's a lot of conflict between students and teachers there, and students say, “Oh, I used it.” [AI], but it's still my job. Then the teacher says, “Any usage is too many.''
Or what is too much and what is too little?
Because any teacher can tell you that it's a delicate balance. Classroom management is a delicate balance. You are always managing your child's emotions, their situation that day, and your own emotions. And you're trying to build trust, maintain trust, and foster trust. We have to keep this very delicate, beautiful and important thing from falling to the ground and breaking into millions of pieces.
Listen to the entire conversation on the EdSurge podcast.
