Why the Push for AI-Era Critical Thinking Can Overlook What Kids Actually Need Most
Third period last week, the heat kicked on too high in my 7th grade digital literacy classroom, and I found Javi slouched in his back row seat, staring blankly at the misidentification exercise on his screen. The worksheet was the standard one we’re all using now: four images, one a deepfake of a local mayor taking a bribe, spot the fake and write three reasons you knew it was AI-generated. Crumbs from a Cheeto bag were scattered across his desk, his Minecraft hoodie was slipping off one shoulder, and when I asked if he needed help, he shrugged and said, “All this is just arguing about what’s fake. None of it matters to me anyway.”
I get why that exercise exists. For the last year, every district training, every education email I get has been focused on the same thing: kids need critical thinking for the AI era. We have to teach them to spot deepfakes, catch biased AI outputs, tell when a student paper is written by a bot. I’ve bought into it, too. It makes sense—AI is everywhere, and kids need to know how to not get tricked. I created that deepfake worksheet myself, even.
Javi stayed after class, I was packing up my laptop, and he hung by the door twisting the strap of his backpack. He said he had a question, not for a grade. Last weekend, his grandma got a text saying her Medicare benefits were getting cut off unless she clicked a link and entered her social security number. He looked at it right away, knew it was a scam—we’d talked about common scam red flags two weeks earlier. “I got that part right,” he said. “But I don’t know what to do now. She’s panicking she’ll lose her insurance when she goes for her knee surgery next month. Who do I even call to make sure she’s okay?”
I froze for a second. I flipped through my unit plan in my head, the 8-week unit on AI and online information. Seven full weeks were spent on how to evaluate what’s real and what’s AI-generated. One 10-minute bullet point at the end of the last week said “discuss next steps for scams.” That was it. We’d never practiced what to actually do after you spot the problem.
This isn’t just a classroom problem, either. Last month, my 10th grade niece came over for dinner, still annoyed about her recent climate change project. She’d used ChatGPT to pull together a first draft of facts about local recycling rates, then did the required assignment: cross-check every fact, point out the three errors and two biases the AI put in the text. She got an A, but that wasn’t what she cared about. She’d wanted to turn the project into a real recycling drive for the school’s parking lot, where kids dump soda bottles and chip bags every day. She had all the facts from the AI, all the practice picking apart its mistakes, but she had no idea how to write an email to the principal asking for bin funding, or how to ask the custodial team to add extra pickups, or how to get other kids to actually use the bins. All the work she’d done for class was about picking apart information, not building something she cared about with it.
The more I think about it, the whole push for AI-era critical thinking has fallen into this trap. It’s all defensive. We’re so focused on protecting kids from the bad stuff AI can bring that we’ve centered everything on evaluation—can you tell what’s wrong? can you spot the fake?—and forgotten to teach the next step: what do you do now? That’s the part kids actually need, the part that matters to their daily lives, whether they’re helping a grandparent with a scam or planning a school project or figuring out how to respond to a fake video a friend shared on TikTok.
It doesn’t take a full curriculum rewrite to shift this, either. I’ve started adding one small question to every lesson now, right after we finish talking about how to spot a problem: “Okay, now what?” If we’re talking about deepfakes, after everyone identifies the fake, I ask them to turn to their partner and walk through what they’d actually do if their mom sent them that same video. Half the time, they first say they’d just tell her it’s fake, but when we push a little more, they come up with concrete steps: send her a link to a fact-checking article from a local news station, help her report the post so it doesn’t get shared more, even walk her through how to block the account that shared it. That’s 5 extra minutes of class, no new worksheets, no extra grading.
For parents, it’s the same small shift. If your kid comes to you saying they got a weird DM or a scam text, don’t stop after asking “do you know this is fake?” Ask them “what do you think we should do next?” and walk through it step by step, even if it’s just practicing what to say when you call a bank to put a fraud alert on a card. If your kid uses AI to help with a school project, don’t only ask them to check for mistakes. Ask them “what part of this can you use for the thing you want to build?” It’s a tiny change, but it shifts the whole focus from avoiding harm to doing something that matters.
I sat with Javi for 20 minutes after school that day. We pulled up the official number for the state Medicare office on the government website, no links from random texts, and Javi practiced what he wanted to say. He was nervous his English wasn’t good enough to explain the problem clearly, so we ran through it twice, with me playing the customer service rep. He called from the hallway outside my classroom, and 10 minutes later he came back grinning: everything was fine, his grandma’s benefits were intact, the rep had noted the scam text and told them to ignore any more like it. A week later he brought me a butter cookie his grandma baked, still warm in a crumpled napkin.
I still teach the deepfake lessons, I still assign the error-checking exercises for AI outputs. Those skills matter, there’s no question about that. But now I add that 5-minute “what next” block, and I’ve started asking kids to bring in problems they’ve actually run into online to work through, instead of only using hypothetical examples from worksheets. Some weeks I forget, some lessons it feels clunky and awkward, and I still have to check all the boxes the district requires. But more and more, kids are raising their hands to talk about things that actually matter to them—scam texts their parents got, weird videos their friends shared, projects they want to do that don’t fit on a test. It’s not a perfect fix, and there’s still a lot we’re all figuring out as AI changes what kids need to learn. But right now, it feels like we’re actually giving them something they can use, when they need it.
Please indicate: Thinking In Educating » Why the Push for AI-Era Critical Thinking Can Overlook What Kids Actually Need Most