While facilitating a PBL 101 workshop at PBL World this summer, I had many conversations that centered on creativity, which was the theme of the conference. One recurring concern was how to create the conditions for students to tap into real creativity in their learning. Most teachers weren't debating whether AI belonged in a project anymore. They wanted to know how to vary its use so a student's original thinking didn't get swallowed by a tool that can generate an idea faster than a kid can finish forming one.
I don't think that's a small question, and I don't think it has a policy answer. Creativity shows up on every list of skills people say students need most, but most schools don't have a real plan for building it on purpose, the way PBL already builds in collaboration or communication by design. That's the case Larmer, Mergendoller, and Boss make in Setting the Standard for Project Based Learning, and it's the same argument PBLWorks's own research has made about digital literacy since the pandemic reshaped classrooms. My April 2026 article, AI in the Room: How Gold Standard PBL Keeps the Learning in the Hands of the Students, looked at process, sustained inquiry, critique and revision, discernment. This post picks up the thread that piece left alone: what happens to a student's imagination along the way.
I've started thinking about it in three moments, the same three moments every project moves through anyway. The launch, when an idea is still raw. The middle, when it's being shaped. And the product, when a student has to stand behind what they made.
The Launch: Protecting the First Instinct
The instinct in a lot of classrooms is to bring AI in early, as a brainstorming partner from day one. I understand the appeal. It's fast and it never runs out of ideas. But speed is exactly the problem. Researchers who study curiosity, going back to George Loewenstein's foundational work in the 1990s, describe it as something that grows out of a gap, the space between what a student already knows and what they're straining to figure out. Curiosity needs that gap to stay open for a while to do its work. A student who asks AI for ten project ideas before they've sat with the question themselves closes the gap before it has a chance to pull them anywhere, and skips the part where their own thinking would have shown up in the first place.
I've started asking teachers to build a no-AI thinking window, even a short one, into the front end of every project.
I've started asking teachers to build a no-AI thinking window, even a short one, into the front end of every project. Before students touch a tech tool, have them sketch, freewrite, or argue it out with a partner. A sophomore group tackling a local history project might spend fifteen minutes debating what story is missing from their town's official narrative, no devices involved. That messy first instinct is worth protecting. AI can enter after that, but as expansion, not origin. A junior working on a public art proposal might write down her instinct first, a mural that changes depending on where you stand, then bring it to AI, not to ask what to make, but to stress-test what she already decided. "Push back on this. What's the weakest part, and what would make it bolder?" The idea stays hers. AI just makes her defend it better. This is really the same instinct behind sustained inquiry in Gold Standard PBL, protecting the gap long enough for a student's own question to take shape before any tool starts answering it for them.
The Middle: Widening What Feels Possible
Left unguided, AI tends to hand back the most statistically likely answer, usually the most predictable one. Ask ten students using the same tool for a project concept and you'll often get variations on the same few ideas back. That's the opposite of what creativity in PBL is supposed to produce, and it's where the middle of a project matters most.
Students need to notice when they've been handed the safe, unexpected answer, and push past it.
This is the point to design milestone checkpoints that ask students to generate a range of ideas before committing to a direction – the same instinct behind student voice & choice as a Gold Standard PBL design element. A student designing a community garden project might prompt AI for three different approaches, one practical, one experimental, one that would surprise someone, rather than asking it to just solve the problem outright. Students need to notice when they've been handed the safe, expected answer, and push past it. A sophomore working on a persuasive campaign might get a generic slogan back and learn to ask, "That's the obvious one. What's a version no one would expect?" That habit is a creative skill in its own right, one students only build if they're taught to recognize mediocrity instead of quietly accepting it.
The Product: Claiming What's Theirs
This is the part that worries teachers most, and it's a fair worry. When an AI tool helped shape language, structure, or a visual along the way, where does a student's authorship actually live by the time the project is finished?
I think the answer has to be visible, not assumed. John Larmer made a useful distinction in PBLWorks's own thinking on this, years before AI entered the picture: you can assess a student's process for generating and refining ideas without ever grading how creative they are as a person. PBLWorks's more recent research on student-centered assessment makes the same case for PBL broadly, that assessment should make thinking visible rather than just measure a finished product. Build authorship tracking into the product itself. Ask students to annotate their final product with brief notes on which choices were theirs and which emerged from AI, and why. A high school student presenting a short film might note the concept and shot list were entirely her own, but that she used AI to troubleshoot a technical editing problem partway through. That's not a confession. It's evidence of the discernment Gold Standard PBL has always tried to build.
Ask students to annotate their final product with brief notes on which choices were theirs and which emerged from AI, and why.
Before final submission, I ask every student two questions: how does your creativity show up or thread through this project, and if you removed everything AI touched, is there still something here that is yours? A student who can point to both has done the creative work. A student who can't is a signal, a milestone check, worth coaching the student through another iteration before presenting their final product.
Creativity Still Belongs to the Student
The teachers I've watched navigate this well aren't running the strictest AI rules. They've built enough structure into the launch, the middle, and the product that students still have to show up with something only they could have made. Protect the first instinct. Widen the field. Make authorship visible instead of assumed.
A project will always look somewhat different once AI is part of the process. What matters most is a student's own reflection, did they own the learning, and can they point to where they left their mark. That's the kind of learning that lasts.
