What My Students Taught Me About AI (That We're Getting Wrong)
Guest post by Jessica Johnson
Earlier this year, I gave my students an unusual amount of freedom.
We had just finished Hour of Code / Hour of AI, and students were curious—really curious—about how AI worked. I had already created a few structured chatbots for them to use (like a Python debugging tutor that helps without giving answers), but during one class, I casually mentioned that I could show them how to build their own.
That was all it took.
I walked them through my general process:
start with background
define tone
add constraints
clarify the role
They were completely locked in.
So I gave them a challenge:
Build a chatbot for anything. We’ll share them in a few days.
If I’m being honest, I expected a very specific outcome.
I thought I’d get a flood of “give me the answer” bots—and that we’d end up having a long conversation about cheating and ethical AI use.
That’s not what happened.
What They Actually Built
Most of my students didn’t try to shortcut learning. They designed bots that supported it.
One 9th grader created Cleopatra, a history assistant with personality and purpose. When asked what she does, she responds:
“Think of me not just as a queen of the Nile, but as your personal architect for the vast library of human history… I am here to ensure you don’t just memorize facts, but truly understand the ‘why’ behind them.”
She built in roles like:
Study Architect
Historical Translator
Critical Thinker
This wasn’t a fact machine—it was a thinking partner.
Another student built a 10th-grade math tutor designed specifically not to give answers. It uses:
step-by-step guidance
Socratic questioning
conceptual explanations
The backend instructions were even more intentional:
never rewrite student code
never give direct solutions
always guide with one question
One student told me:
“I just don’t understand how my teacher explains things… I used the tutor I made and finally understand it.”
He emailed the entire conversation to his teacher—because he wanted to be clear he hadn’t cheated. That moment mattered.
Then there were the creative constraints. One student built a bot that could only respond in Sun Tzu–style quotes. Another created Shmingle the Protector™, a dramatic, over-the-top history guide who repeatedly declared:
“It is I, Shmingle the Protector™. I shall not take up the quill to pen your essays… I am a protector of knowledge, not a ghostwriter!”
Even in humor, the boundary was clear:
help me learn—don’t do it for me.
What This Shifted for Me
We spend a lot of time talking about what students might do with AI. Cheat. Shortcut. Avoid thinking. But when given the chance to design the tool themselves, most of my students did the opposite.
They built:
scaffolds
supports
alternative explanations
systems that respected the learning process
They didn’t just use AI—they defined its role.
AI Isn’t the Problem—Avoidance Is
I hear a lot of reasons to avoid AI in classrooms:
“Students will cheat”
“It replaces thinking”
“It has environmental costs”
Some of these concerns are valid. But avoiding AI entirely doesn’t solve them. It just removes the opportunity to teach students:
when to use it
how to question it
how to verify it
I often tell my students:
“Did you ask where it got that information?”
Because context matters. If you ask AI how many times Jennifer Lopez has been married, you might get different answers depending on the source:
real life: four
film roles: significantly more
Both can be “correct.” Only one answers the actual question.
What I’m Taking Forward
AI isn’t going away. It’s becoming this generation’s version of “just Google it.” And just like with search, the goal isn’t restriction—it’s responsibility.
So instead of asking:
“How do we stop students from using AI?”
I’m asking:
“How do we help them use it well?”
Because if my students are any indication, they’re already trying to.
Try This in Your Classroom
If you’re not sure where to start with AI, this is one of the simplest and most revealing things I’ve done:
1. Show them how to build a chatbot (briefly)
Walk students through:
background (what it knows)
tone (how it sounds)
constraints (what it can’t do)
role (what it’s for)
Keep it short—you’re modeling, not over-teaching.
2. Give them one open challenge
“Create a chatbot that helps someone learn something.”
That’s it.
No long rubric. No over-structuring.
3. Watch what they choose to do
This is the important part. You’ll learn very quickly:
who wants support vs. shortcuts
how they define “help”
what kind of learner they are
4. Share and reflect as a class
Have students:
demo their chatbot
explain their design choices
talk about what they allowed vs. restricted
This is where the real learning happens.
5. Use it as a conversation starter—not a gotcha
Instead of leading with:
“Don’t use AI to cheat”
You can ask:
“What should AI do for you? What should it not do?”
Let them wrestle with it.
Jessica Johnson is a computer science teacher in Boise, Idaho, co-founder of the Idaho Google Educator Group, and creator of “Jessica’s Corner,” where she shares practical edtech tools for the classroom. She is passionate about making tech accessible, creative, and meaningful for all students.
Self-Portrait by Jessica Johnson