AI Brainstorming: Start with a Spark
Last week, I shared how asking smart questions helps kids use AI as a thinking partner. When kids learn to prompt well, they're not just getting better answers—they're practicing clarity, curiosity, and revision.
But AI’s speed and confidence make it tempting to let it do the thinking for us. It’s fast, it feels productive, and it saves time—three things often in short supply in family life. Still, our AI parenting mantra is simple but powerful: thinking first, AI second.
AI as a Spark Partner
The most powerful way kids can use AI to brainstorm isn’t by asking, “Give me an idea for my science project.” It’s by saying, “Here’s my idea—help me make it better.”
When kids bring their own ideas first—a story, invention, or experiment—they flip AI from being an idea factory to a spark partner. Instead of replacing their thinking, AI challenges and refines it.
That shift—thinking first, AI second—isn’t just a parenting hunch; it’s backed by science. MIT researchers recently found that when people rely on AI to generate content from the start, they lose what the study called “perceived ownership” of their work and even struggle to recall what they wrote minutes later. But when they brainstorm first and use AI to build on their ideas, their brains stay more active, and memory and understanding improve.
A Lightning Bolt of Curiosity
A few months ago, my 12-year-old son asked out of nowhere, “Could we use lightning to power cities?”
My first instinct was to explain why it wouldn’t work. But I caught myself—why shut down curiosity just because the idea seemed far-fetched? So instead, we asked ChatGPT together.
AI didn’t just say no. It explained why lightning capture has failed—the unpredictability, the storage problem, the cost—and then pivoted to renewable-energy research that is working. We discovered piezoelectric energy, how motion can create electricity.
That’s the moment I love most: when an “impossible” idea turns into grounded curiosity. And it only happened because he started with his own idea.
What We’re Testing at Home
We’re trying a few new AI-use strategies at home—and learning along the way.
1. Start with Their Thinking
Before AI gets involved, I ask my kids to sketch something, outline something, or explain their idea out loud. It doesn’t have to be perfect—sometimes it’s a messy drawing or a quick brainstorm—but that’s the point. The goal isn’t polish; it’s ownership.
Last week, my 10-year-old daughter wanted to design an order form for her cupcake business. Instead of asking ChatGPT to make it from scratch, we listed what mattered to her—flavors, pickup times, special requests—and then used AI to format and tidy it up. She ended up with something that was hers, just cleaner.
It turns out the best AI brainstorming doesn’t come from giving chatbots total freedom—it comes from giving them structure. A 2025 study from Tel Aviv University found that chatbots generated more original and useful ideas when given clear parameters and focused tasks, rather than open-ended instructions.
2. Frame AI as a “Critical Friend”
A critical friend isn’t harsh; they ask thoughtful, challenging questions. When we use AI this way, it stops being a shortcut and becomes a thinking partner.
Sometimes AI’s best role is to identify what we haven’t considered: Did you think about this? What might go wrong? Asking AI these kinds of questions pushes beyond first-draft thinking. I use this approach in my own work, often asking AI to anticipate the questions or reactions my audience might have. Used this way, AI becomes less a tool that gives answers and more a partner that helps ideas grow.
This approach also helps counter what researchers at the University of Hong Kong call “creative homogeneity.” Their 2025 study found that many large language models tend to produce remarkably similar ideas, even when asked for creative ones. When we bring our perspectives and use AI as a critical friend, we push towards unique thinking that AI can’t replicate on its own.
3. Encourage Back-and-Forth
When my son works with AI, I remind him not to stop at the first answer. I’ll ask, “What else could you try?” or “Does that response really fit what you meant?” Sometimes we’ll go through several rounds before he lands on something that feels right.
That back-and-forth is where the real learning happens. Each exchange teaches him how to evaluate, adapt, and clarify his ideas—skills that matter far beyond the project at hand.
There’s research supporting this iterative approach. A 2025 study found that when AI generated multiple responses—five or ten ideas instead of one—the combined creativity of those responses rivaled what small human teams could produce. But it’s not about asking AI for ten random ideas—it’s about using those rounds to refine, combine, and strengthen your own.
Why This Is Hard (and Why It’s Worth It)
Sure, it would be faster to let my son type “science-fair ideas” into ChatGPT and call it done. It would save time to let my daughter ask AI for her story starter. But speed isn’t the goal. Learning is.
Every time my kids pause to think first, sketch first, or talk through an idea before turning to AI, they’re building something no tool can replicate: the ability to think originally. And honestly, your kids may find, as mine often do, that once they’ve brought their own ideas to the table, brainstorming with AI is not only more productive—it’s a lot more engaging than brainstorming alone.
The AI Brainstorming Challenge: “Idea First” Experiment
Here’s a creative, hands-on way to try AI brainstorming with your kid(s):
Pick a real project your child is working on or wants to start—a book report, a design, a game, anything they care about.
Set a timer for five minutes. Have them capture their thinking however they like: sketch it, jot notes, or make a quick voice memo. No AI yet.
Use AI as their critical friend. Type or say: “Here’s my idea: [their thinking]. What questions should I consider? What might I be missing?”
Go two rounds deeper. After AI responds, ask your child, “Does that fit what you meant? What else do you want to try?” Keep the conversation going until their idea feels sharper, not flatter.
Then, observe. Notice their energy—are they defending their ideas, revising, or exploring new directions? Their engagement in creating something, not the final product, is the goal.
What's Next
In my next post, we'll tackle something I've been wrestling with: AI sounds confident even when it's wrong. So how do we teach kids to fact-check without turning research into a chore? I'll share a simple 30-second habit called "lateral reading" and a few other strategies that help kids verify what they're learning—without needing to become professional researchers.
If you liked this post, subscribe below and share it with friends!
Sources
Kosmyna, Nataliya, et al. “Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing Task.” arXiv, 2025.
Mizrahi, Alon, et al. “Cooking Up Creativity: A Cognitively-Inspired Approach for Enhancing LLM Creativity through Structured Representations.” arXiv, 2025.
Korb, Kevin B., et al. “Large Language Models Show Both Individual and Collective Creativity.” Journal of Computational Creativity, 2025.
Zhu, Ling, et al. “Creative Homogeneity Across Large Language Models.” arXiv, 2025.
