My ChatGPT and Generative AI Policy This Year
Like many teachers, I spent a lot of time this summer thinking about how I would handle AI in my classes. That journey was often an all-seasons-in-a-day experience. I would go from excitement over its possibilities to existential fear concerning AI’s potential implications to concerns about whether allowing students to farm out planning or proofreading to a machine would cause their abilities to do those things to wither on the vine.
In this process, I must have drafted a dozen AI policies for my class and AI posts for this newsletter, but none fit. It wasn’t until a first week of school gathering that I finally had a breakthrough while talking to a fellow teacher about a successful experiment she did in her first week. This teacher expressed that she’d long been frustrated by how long it took for students to quiet after the bell rang. Over the years she’d tried a number of approaches to help them quickly settle, but none worked as well as she wanted.
This year, based on inspiration from a French class that had a call and response opening where the teacher and students would say good morning/afternoon in French each day, she decided to do a call and response too, but she found herself struggling with what kind of call and response makes sense for history classes. During her first first day of class though she had one of those spontaneous and serendipitous teaching moments where she simply pondered out loud to the class what sort of opening call made sense for history. And that was all it took for her students—in the endlessly brilliant way of children—to start creating one hilarious opening chant after another.
Now when she walks into her room, she says one of the three finalists (as crafted and voted on by the students) and they then give their response:
Her Final Three
Option #1: She says, “Quick, Andy Is Coming,” (a reference to Toy Story) and the students yell “Hide!”
Option #2: She says, “A hush fell over the crowd,” and the students chant, “Hushhhhhh.”
Option #3: She simply says, “Paw Patrol,” and the students gleefully retort, “On the double!”
She reported to me that the results have been like magic. After the communal incantation, which adds to the camaraderie of class as well, the students are instantly quiet and attentive, ready to go with the day’s lesson.
As I talked with her, I suddenly knew how I would approach AI in my classes. In recent years I have engaged students on a regular basis to figure out solutions for thorny problems or quandaries we face in the classroom. Key areas where I’ve consulted with students include how to approach grades, what mentor texts to use, and how to make poetry more engaging (it ended in the inclusion of Shel Silverstein in my curriculum). Talking to this teacher helped me to realize that co-creating an approach with students is perfect for dealing with AI because it is so big, so new, and shifting so rapidly. All summer long I kept trying to be the expert—to figure out exactly how my students should or shouldn’t use AI. But in that moment I decided to create a lesson where I opened up discussion of AI and what we should do about it to the students. Here is what that looked like:
The first step was to introduce AI to students. I found most had heard of ChatGPT and AI and maybe a little over half had actually used or played with some form of AI, but there were a few for whom this seemed like actual news. So I began by giving the basics about how these bots work: They’ve been fed billions of lines of text, they don’t know what they are saying and instead are basically really good guessing machines, and even if students don’t know it, bots are being embedded into everything from Snap to Grammarly to Google.
We then took AI for a spin so they could see it. I pulled up ChatGPT and asked the students for prompts. As usual, their ideas were glorious: a detective noir about a bee searching for the lost pollen, a tale of a dragon the size of a water bottle, a hot dog who wins a hot dog eating contest (although the bot did not like the ironic cannibalism approach and reprimanded us). After we read each story, we talked about them, and the students saw it as clearly as I did: While each started out moderately interesting, the stories quickly settled into cliches, an oddly vacant style, and predictable (and dull) story arcs.
I then shared my essays of the week from last week: Excerpts from two short essays from Margaret Atwood and Stephen King talking about how they felt when they learned that pirated copies of their books were used to train AI.
In true Essay of the Week style, the students took some time to respond to these articles and what they noticed in the stories we had ChatGPT write. The conversations afterwards were amazing, and as we talked, I made sure to ask three key questions:
Are these AI chatbots ethical, given that they were trained on the work (and some of it pirated work at that) of real people?
How do they feel about the bots, both in terms of their current abilities and future possibilities for good and bad?
What role should AI play in a classroom in 2023-2024? Are there legitimate uses or should it be banned outright?
And, like with my friend whose students invented amazing classroom opening chants, the conclusions reached by my students were so much better than my false starts this summer.
The next day I then brought in a policy based on what we discussed. In essence, it was simply an invitation to continue the conversation:
My Policy
Generative AI is a new and rapidly evolving tool. As such, you must discuss the ways you are thinking of using AI (if you are thinking of using it) with me in advance of using it. If you use it (after talking with me), you will also be required to document and share how you used it with me.
The main debate concerning AI in the classroom as I understand it is about how we can best prepare students for the brave new world they will inhabit. Should we be teaching them how to use these tools, which are both powerful and not going away? Should we focus on building skills first so they can better drive these tools later? Or does the truth lie somewhere in between?
I have read and thought about this a lot, but even still, I keep bumping up against hard limits to my understanding of the issue and my ability to predict how it will affect the future. This is why my policy for the first semester is one of curious discussion instead of hard dictates. In a sense, I am ironically doing what generative AI strives to do: Inviting in the thinking and experiences of everyone—students, other teachers, parents/guardians/caretakers, support staff, etc.—in the classroom to ideally create something greater than the whole.
I should end by saying that this doesn’t mean I will allow my students to use AI in whatever ways they want. I am deeply concerned about how outsourcing parts of the writing process could affect student skill development and the moral implications of using something fed on the work of others who were not compensated. Instead, I am allowing students and those who support them to think about these problems too and open up a dialogue so we can learn to navigate the shifting and unclear labyrinth of this new technology together. In that spirit, if you have an approach to AI that you think might help students, please share it here and I can pass it along!
Thanks for reading and yours in teaching,
Matt
One Last Reminder!
My first cohort of my Better, Faster Feedback course closes this Friday. There will be other cohorts, but here are a few reasons to sign up for it now:
I am offering a subscriber discount of 50% for this cohort only (use the code SUBSCRIBER)! Also, if you don't have time right now to do it, once you sign up, you will have access to it for the entire year.
The course is three hours, but the ideas and materials in it will hopefully pay back that time with handsome interest.
If money is an issue, let me know. There are lots of potential funding streams—from PTSOs to professional development funds—and I can offer scholarships to those in need as well.
And for those who signed up already, look for a welcome email on Friday with information about the live sessions and access to my quarterly feedback review.