PRESQUE ISLE, Maine — Generative Artificial Intelligence is a powerful new tool that has caused a lot of debate about its use in school classrooms.
Presque Isle schools Curriculum Director Jennifer Bourassa talked about AI’s potential uses and problems for educators during an SAD 1 board meeting this week, and ended with recommending teachers communicate with an AI entity known as ChatGPT to better understand how it works.
ChatGPT enables a user to have text-based conversations with it and generate emails, essays, and other written material. It does this by mining data on the Internet. It could be used by students to generate school work. Plagiarism is harder to detect when using AI to write essays because the large language model on which it is based is drawing from millions of different documents online.
Bourassa said SAD1 is going to help high school students learn how to use generative AI in the classroom and the ethics of using ChatGPT rather than ban its use and that of similar systems.
“We want to know that kids are using AI effectively as well as honestly and they are doing their own work when we ask them to do their own work,” Bourassa said.
She doesn’t believe banning AI in the classroom is a reasonable response for schools because it is already being used in phones and apps, which most students use, and there are educational benefits. Students can use ChatGPT as a tutor to get feedback on papers to improve their writing, or teachers can generate and enhance a lesson plan.
One example Bourassa talked about was when she tested an old student’s graded paper with ChatGPT. The AI was able to grade the paper and give it well-written feedback in five seconds.
One problem Bourassa pointed out for educators is how to know if a student is using ChatGPT to do work for them.
If a student knows they can’t write perfect essays they can give ChatGPT a prompt to create an imperfect paper that mimics how they would have written it, making cheating harder to detect, Bourassa said.
“In some ways it’s like an automatic plagiarism, which is what some people call it,” Bourassa said.
An update to the SAD1 plagiarism policy will be needed since it’s grounded in the 1980s, she said.
Teachers should ask their students to write papers by hand as one way of getting around ChatGPT in the classroom and see what each student’s writing actually looks like, Bourassa said.
“[ChatGPT] is improving very rapidly and it is probably going to be what we think of as a disruptive technology,” Bourassa said.
She compared ChatGPT to the iPhone when it first came out in 2007. Bourassa pointed out that other apps already use a version of AI technology like Google Maps, Youtube, Siri, and Netflix.
Another problem she outlined is when AI hallucinates, or makes up information to a question it doesn’t know the answer to, along with citations and resources that are false.
Educators at SAD1 need to better understand generative AI and its limitations so it can be taught to students, Bourassa said.
So far, SAD1’s high school English department is using prompts with ChatGPT with high school students to better understand where the AI can fall down, although a structured lesson plan is not in place.
