What I Learned About AI in Education As a Journalist and Parent
The back row of Ms. Lopez’s 10th grade English class smelled like lemon disinfectant and unwashed gym bags last month, when I was there shadowing for an education feature I was writing. My coffee had gone lukewarm an hour in, and I was scribbling notes in the margin of my notebook when the boy sitting next to me, Javi, tilted his screen toward me. “Ms. Lopez said it’s okay to show you,” he said, tapping the open document. It was a personal narrative about his grandma moving into a memory care facility, and the first line – the one he’d ended up with – was “Her hands still remember how to braid my hair even when she doesn’t remember my name.”
I asked him how long he’d worked on that opening. He said he’d messed around with six bad ones before he even opened ChatGPT, then he asked AI to give him three more options. “All of AI’s were cheesy,” he said, “but it made me realize I wanted to talk about her hands, not how she forgot my birthday. So that’s where I started.”
I’d gone into that classroom ready to see what I’d read about in all the breaking headlines: kids cheating en masse, teachers scrambling to catch up, a whole generation losing the ability to think for themselves. I’ve been covering education for five years as a journalist, and for the last 18 months, AI has been the only story anyone wants to talk about. But that afternoon with Javi and Ms. Lopez, and then a fight with my own 7th grade son that night, upended most of what I thought I knew.
After class, Ms. Lopez leaned against her desk and ate a granola bar while we talked. She said she doesn’t ban AI in her class, but she requires every student to turn in a “process note” with every out-of-class assignment. The note just needs two things: what part of the work AI did, and what part they did themselves. “If a kid says AI wrote the whole essay and they just put their name on it, I give them a zero and we talk about why that’s not helpful,” she said. “But if they say AI cut 200 words of fluff or helped them work through a clunky paragraph, that’s fine. That’s what I do when I’m working on a lesson plan, anyway.”
That conversation was still bouncing around my head when I got home that night, and found an email from my son Leo’s math teacher waiting in my inbox. Leo had turned in an extra credit project on how exponential growth works for social media followers, and the school’s AI detection tool had flagged it as 92% AI-generated. The email asked me to come in for a conference if he’d cheated.
I found Leo at the kitchen table, eating cold pepperoni pizza and scrolling TikTok, and showed him the email. He rolled his eyes, grabbed his laptop, and pulled up a folder full of messy handwritten notes, screenshots of his own follower count over two weeks, and the chat log with AI. “I did all the data myself,” he said, pointing to the notes. “You know how long it takes me to write a paragraph because my dyslexia makes me mix up the words. I just asked AI to put my notes into full sentences so I didn’t get points off for typos and run-ons. That’s not cheating.”
He was right. I’d forgotten that for kids with learning differences, AI isn’t a cheating shortcut – it’s an accessibility tool, the same way a calculator is for kids who struggle with basic math, or audio books are for kids who can’t read as fast as their peers. For years, Leo’s grades have never matched what he actually knows, because he loses points for messy writing and spelling errors. Now that he can use AI to clean up his work, his teacher is finally seeing how much he understands.
After that, I started asking every teacher and parent I talked to for my stories what they actually do about AI, not what they think they should do. The good solutions I found aren’t the fancy, expensive ones edtech companies sell, and they don’t involve banning AI from classrooms entirely.
The first thing that works is just being clear about expectations. Most of the teachers I talked to who have no major AI problems split assignments into two categories: skill-building work done in class, no AI allowed, because the point is to practice the skill itself. For example, in-class essays are where you practice organizing your thoughts under pressure, so no AI. Out-of-class projects that are meant to show you understand a concept can use AI for the busy work – formatting citations, cleaning up writing, brainstorming ideas. That’s concrete, easy for both teachers and parents to follow.
The second thing is stop asking kids “did you use AI” and start asking “what did AI do for this work, and what did you do?” I started doing this with Leo, instead of immediately assuming he cheated when a teacher flags something. It takes two minutes, and it stops a lot of unnecessary fights. It also teaches kids to be transparent about how they work, instead of hiding what they’re doing.
The third thing that comes up over and over is that AI detection tools are wildly inconsistent. Half the time they flag work written by kids with dyslexia who use AI to clean up their writing, and they miss work that’s been heavily edited by AI anyway. Most teachers I know stopped using them after a few months, because they caused more problems than they solved. Instead, if a teacher thinks a kid didn’t do the work themselves, they just pull them aside for a five minute chat and ask them a couple of basic questions about their project. If they can’t explain it, they get to redo it. That’s way more accurate and way less stressful for everyone.
I still have questions, of course. Last week I sat in on an 8th grade history class where a boy turned in a perfect AI-written essay on the Montgomery Bus Boycott, but couldn’t tell me who Rosa Parks was when I asked him after class. That’s a real problem, and it doesn’t get fixed just by being transparent. Some days I worry that Leo will get too used to having AI structure his paragraphs, and he’ll never get good at organizing his own thoughts on paper.
Last night, he asked me if he could use AI to practice his Spanish pronunciation for his quiz next week. I said yes, but we sat down and went through all the vocabulary flashcards together first, before he turned AI on. It’s not a perfect system, and I don’t have all the answers. But after a year of talking to teachers and parents and kids, I know the panic I felt a year ago – that AI would break education entirely – was overblown. It’s just a new tool, and we’re all still figuring out how to use it right.
Please indicate: Thinking In Educating » What I Learned About AI in Education As a Journalist and Parent