A Primer for Returning to the Classroom with AI
Here's a compilation of stuff I think you need to know
A new academic year is dawning, at least in the South, where we start the year early and end it early. My college semester starts August 18, but some of the K-12 schools are starting as early as July 25th (yes, tomorrow). So, I better get busy with writing my “Back to School with AI” blog.
I’m very excited because, when I return to the classroom for Fall Semester 2025, I can say authoritatively that I have been teaching with AI in my composition courses, in almost every lesson, FOR TWO FULL YEARS!! (Pause for applause). That makes me one of the earliest adopters in the nation, if not the world (the other earliest adopter would be my dear compatriot,
.) So, I have a little experience doing the practical, everyday work of teaching with this amazing, scary, frustrating, magical, overwhelming tool. But let’s get to the good stuff.I wanted to make you an official Fall 2025 Back to School list of things you should know when you teach with AI. I have written about a lot of this already, so if you are a regular reader, please bear with me.
New Definitions for Old Terms
Teaching will never be the same. What you were doing last year or the year before needs to be rethought, rearranged, or completely redone. (Take a moment of silence to mourn your old, dead syllabus. It’s OK. I’ll wait.). 🕛🕧🕐🕜🕑🕝🕒
Here’s some helpful definitions:
Friction : What we need to ensure critical thinking is happening.
Learning should be hard, it should be a struggle. Students shouldn’t just skate along with no friction because friction means they have to slow down and figure things out. Always aim to add friction to your assignments whenever possible.
Cheating: Using AI to avoid friction in learning.
If you are wondering whether a student is cheating with AI, you can apply the question, “Did the student use AI to avoid friction?” Cheating is a normal human response to friction, but it is not necessarily an ethical or moral response. There will always be cheaters, but cheating isn’t because of AI. Sometimes cheating with AI is not malicious. It stems, instead, from one of two reasons: the student doesn’t know how to use it ethically or responsibly because they haven’t been taught how, or students think they will “never be as good as AI, so why bother trying?” Before you condemn a student for using AI inappropriately, please make sure you understand why they did it. This is what smarmy pop-psych people like to call a “teachable moment.” Hey, wait! You’re a teacher! How convenient is that? So, take off the damn policeman’s hat, put down the whistle, and teach!! (The crowd goes wild! Yeah!)
Create
Remember when “create” was at the top of Bloom’s Taxonomy? Of course you do! (Bless your heart.) Now, it’s at the bottom because when you teach with AI, students create first. So, that means you need to work your way backwards through Bloom’s Taxonomy to teach your students to Evaluate their creation, Analyze their creation, Apply their creation to new situations, Understand their creation, and finally Remember what they have done to understand the implications of their creation. I know, I know. I just turned your world upside down 🙃. It’s OK. I was the one who reimagined it this way, and it took me a while to wrap MY head around it—because I created it first, then I went back to evaluate, analyze, apply, understand, and finally remember what I did. It’s kind of complex, so I wrote a lot about it here and also here. (I will probably write a lot more about it soon as well.)
Some Basic Best Practices
I’m going to write this section as if you are a newbie to teaching with AI and you don’t know much about it. If you are a newbie, that’s OK. I’m glad you are here reading this now. Some of us are crazy and ran at the whole AI thing the moment it came out like some possessed Don Quixote (ahem). Some are more moderate and slowly warmed up to it, adopting it little by little. Then, there are some who are possessed of wisdom and patience who, at first, tried to wish it all away, then ignored it for a time, and are now curious about what to do with this newfangled thing. So, here’s some established best practice stuff you might want to adopt:
Syllabus Statements
This is the most basic thing that has come out. I’m sure your department chair has already asked everyone to do one of these. You may have started out completely banning AI, then realized that was impossible. Now, you want to make a new statement that is more encompassing and less restrictive, but you don’t know where to start. Here is a great resource from
, who collected a whole bunch of syllabus statements from everywhere:
Don’t Use “AI Detectors,” Ever
Not only are AI detectors very unreliable (try running your old papers through one!), but they also present a huge problem in your classroom. You should not be putting your students under suspicion of cheating just because AI exists. It is the opposite of establishing a trusting, caring, and open environment for learning. Once you use an AI “detector” even once in your classroom, you will sow fear and resentment in your students. Just don’t do it. Here’s why I don’t use an AI detector, and why you shouldn’t either. The “teaching with AI” community is in nearly 100% full-throated opposition to AI detectors. Please don’t use them. They are horrible.
Teach Ethical AI Use
Remember up at the top of this article when I mentioned that a lot of your students won’t know how to ethically use AI? Well, now you are down here, and I am telling you that this is an important aspect of everything you do in your classroom from now on. There are a few important things they need to know:
Transparency statements—they must always say what, how, and why they used AI (or didn’t) in every assignment they do. We want to make them very aware of their AI use. I require students to put the transparency statement at the top of the paper, right after the title.
Define what “cheating” is in your classroom, and be very transparent about when they can and can’t use AI for your assignments and why. I’m not talking about your syllabus statement here. I’m talking about on the assignment level. Make it very plain in every assignment—you can use AI here, you can’t use AI there, etc. Explain why you have chosen to restrict it sometimes and how that helps them learn.
Model ethical AI use. Pay attention to how you use AI. Give transparency statements when you use it to make an assignment, an image, or even grade work. Show them that you are using it responsibly.
Open your classroom to the AI conversation. Let students vent. They have the same fears and worries and excitement and thrill that we have. Talk about it. Let them be honest with you, and you should be honest with them. If you aren’t comfortable using something, tell them about it and why. Ask them for advice, and share your frustrations, your challenges, and your excitement. They need to talk this out with someone who is knowledgeable and caring so they can understand what is going on.
Study up on how AI works, what the security issues are, and what students and faculty need to do to protect their privacy and their data. This is a great opportunity to get students researching how AI privacy and security works. This is important information for them to understand and share. Take a look at this really great overview of AI Privacy and Schools for some great places to start with this knowledge base:
Understand what AI literacy is and how it relates to you and your students
On the most basic level, AI Literacy is a fundamental understanding of how AI works and how to use it. There are some excellent AI Literacy Guidelines out there, including the one I wrote for EDUCAUSE. Check with your school librarian to see if there are any that have been specifically adopted to your school, district, or state. If not, you might want to do a little research into some literacy guidelines that might apply in your specific case depending upon what subject you teach and at what level. Here’s a great article by
to get you started:
Finally, some AI Pedagogy Tips
and I have worked very hard at developing some AI Pedagogy that works very well in our writing classrooms, and a lot of it can be transferred over to any subject. The biggest thing to remember is that AI didn’t change education; it simply exposed where we were doing it wrong. Remember all those times when the education department people came to your department and tried to convince you to carefully scaffold your assignments, or use formative assessment, and you ignored them? Yeah. That’s the exposure I’m talking about!Rework your Assignments
It used to be OK to write something as simple as “Research the fall of the Roman Empire and give me a summary of how it happened.” AI can do that without breaking a sweat, and your students know it. You need to be more specific, and you need to be asking for a lot more from your students. There are a few things that we have found to be helpful in keeping the student engaged in completing assignments.
Formative over Summative assignments. You need to have a lot of short assignments that let you know how the students are doing along the way. The days of having one big essay exam or a giant research paper at the end are gone. You need to be checking on authentic learning all along the way.
Scaffold your assignments. Every assignment should build a skill for the next assignment—leading up to that big assignment at the end. They all should count for the same amount (or close to it). If you did ten assignments to get to the research paper, then they are just combined for that paper. You don’t need to give it more points. If you have trouble figuring out how to scaffold assignments, put your big assignment into your friendly neighborhood chatbot and let it help you do it.
Metacognative Reflection is essential. Eugenia Novokshanova invented the “reflective footnotes” that we use in our classes. Every paper they write must have at least ten reflective footnotes that talk about their process, why they made the choices they did, and how they felt about those choices. This can be adapted to anything. For example—lets say you teach calculus. You can require that students explain why they used a particular formula over another to solve an equasion. This helps you understand where the student may have difficiencies and allow you to address them immediately. It also helps us figure out how to teach the next assignment more specifically to what they need to know.
Ideas and concepts over facts. Don’t ask simple questions about what they should know. That is so easy to fake. Make them apply knowledge, think big, do something creative. In literature, I asked my students to generate a picture of a modern “Wife of Bath.” What would she look like? How did they match the description in the Cantebury Tales? How was her look updated? What do you think she would do for a living today? Why do you think that? This forces students to look deeply at the text and try to find clues beyond simple facts.
Integrate multimedia items. Multimedia is easy now. Students can make pictures, videos, infographics easily and quickly. Teach them how to integrate multimedia elements into what they do. Leverage AI to teach with multimedia elements. For example: Ask the student to tell AI to make a picture of a plant cell. Then, challenge the students to find three things that are right and three things that are wrong with the image.
In writing assignments, tell students you must be able to differentiate their voice from the voice of AI. If you can’t tell the difference, they will need to revise their work.
Tell them they have to write for an audience of their peers. This means the five-dollar words that AI spews won’t fit here.
Keep the focus on voice. Make them read their work aloud to another student. If they are online, have them make a video of them reading their assignment aloud and share it with another student. Explain that they need to develop their voice. Any words they can’t say, they should rewrite. Any phrases that don’t work in their mouth need to be reworked.
Don’t go “full monty” in your assignment sheets.
I know you were taught to put everything on the sheet, and make it as clear as possible. When you do that today, you end up with your students feeding the assignment sheet to a chatbot and playing “push button get assignment.” Stop doing this! Don’t let it all hang out—be alluring and diffuse.
Give very specific instructions: When we do an assignment, we give very specific instructions for the structure, formatting, and requirements for the paper. For example, we require they give us a transparency statement, a rhetorical image, a unique title. We want the first paragraph to be a specific way, the 3-5 paragraphs to be very specific, the fourth paragraph to have a certain element, etc. Not only does this help students learn to follow directions, but it also confuses the heck out of the chatbot.
Refer to Content Pages: Say things like: “Please see to the content page about integrating quotes for specific instructions.” You can also use links to content pages instead of completely explaining things on the assignment sheet.
Invent some eccentric terms: Rename a common process or a way you want the to do something, then use it in your assignment sheet. For example, instead of telling students to write their paragraphs in a PEEL format (a very common term that AI understands), we renamed it the CLEAR format and added one more step. The AI doesn’t know what we are talking about, but the student knows.
Rethink your Assessment
We realized early on that there were a few important skills students would need to develop in order to survive in the Age of AI. Here are some of the things we found to be important:
Follow Directions
Sound Human
Transparency and Ethics
Check Facts
Be efficient
The next thing we did was adapt those into our rubrics. At this point, we expect our students’ papers to be flawless in spelling, grammar, and mechanics. So, we took that off of our rubric. We substituted the following:
They followed all the directions. Everything was correctly done and correctly formatted.
They payed attention to voice, tone, and audience awareness.
They had a transparency statement and metacognative reflections discussing their process.
Everything was factual and backed up with real evidence that actually exists.
Their work showed excellent use of AI tools such as image generators, tutors, and editors.
Well, those are some of the basics. Please let me know if I forgot anything! (I probably did), or give me some suggestions in the comments!!
Thank you for this primer! I'm sharing it far and wide with my colleagues at St. Petersburg College in Florida.
Very nice series of practical recommendations.