I have spent the past year and change exploring the possibilities, limitations, and risks of Large-Language-Model AI, especially ChatGPT. In my role at work, I haven’t done a lot on it because we don’t have an adopted Board Policy on it, and TSBA hasn’t written a model policy for boards. This leaves us in a weird space where we know that this is out there and people are using it, but we don’t have any guidance or governance for it. I don’t think ignoring it for now is the answer, and I wanted to share what I have communicated so far so that it might be helpful to other districts.
Handbook Policy
You don’t need a board policy to have a handbook policy, so we put this in our high school handbooks at the beginning of the year:
AI Handbook Entry for Academic Integrity and Honesty in the Use of Large Language Models
· Purpose
This handbook entry aims to ensure the upholding of academic integrity and honesty in the context of the use of large language models (LLMs) such as ChatGPT in our school environment.
· Scope
This handbook entry covers all students, staff, and any other individuals who interact with our school's academic programs and services, and who use LLMs for academic purposes.
· Handbook entry Guidelines
Proper Citation: Students must properly acknowledge and cite the use of LLMs in their work. Any idea, phrase, or output generated by AI must be cited just as any other source would be.
Original Work: While LLMs can be used for assistance and guidance, the work submitted by students must fundamentally be their own. The use of AI should be to facilitate and enhance the learning process, not to replace individual effort and creativity.
Collaboration: While working collaboratively, students must clearly state the contributions made by AI. Collective work should reflect a clear understanding of the contributions made by each student and the AI model used.
Access: All students should have equitable access to AI tools to ensure fairness. The school will strive to provide the necessary resources and training for all students.
Educator Guidelines: Teachers should educate students about the ethical use of AI and its potential impacts on academic integrity. They should also receive regular training to stay updated on the capabilities and limitations of AI.
· Implementation and Compliance
This handbook entry should be communicated effectively to all relevant parties. The school will conduct regular checks to ensure compliance. Any violation of this handbook entry will be considered a breach of academic integrity, and the school's standard disciplinary measures will be applied.
Simply enough, we have let students know ahead of time that students can’t use LLMs to produce final products. We’ve also let teachers know that they need to be teaching students how to use LLMs to their advantage.
Where are we now?
We haven’t, to my knowledge, had any issues with students getting caught cheating with LLMs, but that doesn’t mean it hasn’t happened. In fact, the whole inspiration for me writing this is that a student told me that she wouldn’t use AI to help her study French because another student had submitted an essay in her English class using AI and she got the same grade as him and it made her angry.
Because of that conversation, I put together a document for teachers, and I thought I’d share that content here.
So why not just avoid AI for as long as we can?
· You can tell when AI has written something, and I’m surprised when anyone can’t. Have you used AI enough to pick up its tone and patterns? It uses too many adverbs. In emails, it always says some affectation like “I hope this email finds you well.”
· AI isn’t going anywhere. As a matter of fact, it’s the worst quality and the least integrated today than it will ever be in our students’ lives. We have to learn to live with it, and students are going to need to know how to interact with AI now. It can really give them a huge advantage in life if used ethically and responsibly.
· Withholding the power of any technology from our students only withholds it from certain students. Typically, only the students who are disadvantaged will not learn to use technology when it is withheld from them in school.
· We can’t have students using this technology to cheat, and avoiding teaching them how to use it responsibly will not prevent them from cheating. In fact, letting students know that we are very knowledgeable about it will make them think twice about using it to cheat.
So how should students be learning to use AI?
· Helping them get organized.
· Asking it simple questions and interacting with it. For example, this student is having trouble with conversational French. It can have a conversation with her, and she can practice her French with it. You can’t get that anywhere else without a pen pal or French friend.
· Asking it to make a study guide.
· Asking it to quiz you on something.
· Asking it to help you with the pre-writing phase of writing.
· Asking it to proofread your paper (that you wrote) and give you feedback on it. You could even ask it to evaluate the paper with a rubric that the teacher gave them.
· Asking it to explain difficult concepts in simple ways.
· And many other ways…
Here are some samples:
Example: Helping them get organized.
Sample Prompt: We’re learning about cellular energy in my high school biology class in Tennessee. Can you help me get organized with an outline? I will keep you posted on what we’re studying in class so you can help me make a study guide.
Example: Asking it to quiz them on something.
Sample Prompt: We’re studying slope in Algebra I in Tennessee, can you give me some quiz questions and tell me how I did?
Example: Asking it to help with the pre-writing phase of writing.
Sample Prompt: I am writing a research paper on Romeo and Juliet and comparing it to other famous family feuds in more recent history. We’re going to the library to do research next week, and I need to get organized. Can you give me a checklist of what I should be searching? Do you know of any feuds I can research?
Example: Asking it to help with brainstorming
Sample Prompt: In US History, our teacher has asked us to explore the causes of war leading up to World War I. We are supposed to represent a country and their point of view. Help us brainstorm some ideas for this. We can’t choose Germany, Britain, France, or the US. We don’t know these other countries as well. What information do you need to help us with this?
Example: Ask it to proofread your paper and give you feedback.
Sample Prompt: I’m writing a paper for my World History class on the Ming Dynasty, but I need someone to proofread it for me. Can you proofread this and give me a list of suggestions for improving it. Please do not rewrite the paper for me; I do not want to get accused of cheating.
How do I stop cheating?
· Consider whether your assignments are easy for students to cheat on using AI.
· Get experienced enough with AI that you can spot how it writes.
· Take a writing sample at the beginning of the year for a comparison.
· Let students know that you won’t tolerate them using AI for final products, but you’d love for them to use it for brainstorming, outlining, and pre-writing.
I’d love to have a deeper conversation about this, but I want to be clear that we must tackle this issue head-on, and at some point, we’re all going to have to accept that AI is a technology tool that our students need to know how to use. Just like we teach students to use TI-85 calculators, nail guns, MIG and TIG welders, and 3D printers, we have to expose students to all technologies that will help them be successful in life.
There are many AI tools other than ChatGPT that are meant specifically for the classroom. I’m trying to keep a list of them: https://www.jasonhorne.org/ai-tools