How teachers are fighting AI cheating with handwritten work, oral tests, and AI | The machines are winning the classroom

From the article: The fear that generative AI tools such as ChatGPT would lead to a generation of students cheating and plagiarizing work has come to pass. The situation is so bad that educators are now looking at multipe ways to stop the problem, or at least make the practice much more difficult. Ironically, one of them is to use AI.

Speaking about AI-cheat students, Gary Ward, a teacher at Brookes Westshore High School in Victoria, British Columbia, told Business Insider, “Some of the ones that I see using it all the time – I think if it wasn’t there, they would just sit there looking blindly into space.”

There were warnings about AI cheating being endemic in education last year. Now, Ward says that “literally” all students are doing it.

One of the ways Ward is trying to combat the problem is to turn the AI against the cheaters. He asks ChatGPT to help him develop work that would be difficult for students to complete by simply feeding it into a large language model.

Richard Griffin, a lecturer in the business faculty at Manchester Metropolitan University in Manchester, England, is also using AI to make life harder for the AI cheats. The University has developed an in-house system that can be fed assignments. The system will then summarize how difficult it would be to use AI to complete the work, and recommend ways to make doing so more challenging.

There was a good reason we moved away from pure exams and embraced coursework, projects, and other flexible forms of assessment, especially to support different learning styles and reflect real-world skills. However, that model is cracking under the weight of the internet and now AI. It has been increasingly open to abuse since the ubiquity of the internet.

Over the past decade, the potential for abuse and misrepresentation of student work has skyrocketed. And now, with generative AI tools widely available, students can produce “original” assignments with almost no effort or understanding. The barrier to submitting plausible-looking work is basically gone. This is especially true at the high school level where the depth of research and analytical skills is less.

At this point, we may need to take a step back and consider that the only reliable way to assess what a student actually knows may be through in-person, supervised testing or live presentations. It’s not ideal for everything, but at least it ensures the person being assessed is the one doing the work. Otherwise, we risk turning education into a credential mill for those who are best at prompt engineering.

I wonder how many teachers are running essays though GPT to correct them. Basically AI teaching AI through human drones.

The problem here is that until every child has access to an AI assistant the kids that do are going to vastly outperform the ones that don’t.

So make sure everybody has access to one.

And then what you do is you raise the standards accordingly.

It’s like, Okay Little Jimmy, you want AI to write your essay? Then it better be *extremely fucking good*. Like, I would be expecting the definitive thesis on what happens when you give a moose a muffin.

The problem is the teachers who don’t know how to use AI, not the kids who do. 

I say it every time: School is not about learning, it’s about passing specific, time-based performance-checks. In the past we used Ritalin and bullimia-studying to pass those, now the kids use AI, but the core problem is exactly the same!

AI will be what forces educators world-wide to rethink our entire education system and how we approach schooling.

Just do in person exams for grading like the rest of the world? This is really not much of an issue in places where having exams year round is the norm. The solution is obvious and we know it works.

Workplaces are using more and more AI and schools are trying to prevent it. Why not use AI, teach kids to use it, but also to think critically on what AI has produced.