Navigating productive struggle with AI.
Productive struggle is the sweet spot where learners are challenged enough to build new thinking, but not so overwhelmed that they shut down. The goal is to use AI as a scaffold, not an elevator that skips the learning.
Think of the AI as a climbing partner. It doesn't pull you up the rock face, but it can hold the rope, point out a handhold you missed, or suggest a different route. You still have to do the climbing.
The hard part is working out where the productive struggle is, and when. Sometimes brainstorming is where the good work happens and offloading it to AI costs you something real. Sometimes summarizing is the thinking, and doing it yourself is the point. Other times those are just small stops on the way to a bigger problem worth wrestling with.
Using AI to improve productive struggle
Lower the cost of curiosity
Struggle often stalls on the fear of being wrong. AI experimentation is fast and cheap: a team can try five wild ideas in ten minutes. That encourages a playful, curious posture instead of a fearful, perfectionist one.
Provide scaffolding, not answers
A good prompt asks for a structure, not a solution. Asking for the Pixar story spine gives the team a frame they still have to fill, which focuses the struggle on the storytelling rather than on the formatting.
Offer just-in-time prompts for reflection
When a team is stuck, have them ask the AI to help them reflect rather than to solve it: "We're stuck. Act as a facilitator and ask us three questions that would help us re-evaluate our problem statement from a different angle." That deepens their thinking instead of replacing it.
Modeling when not to use AI
To keep learning and ownership genuine, facilitators should model AI off-ramps: moments where the struggle is too important to outsource.
The initial empathy moment
When first reading a persona or listening to an interview, do it without AI. The goal is unmediated human connection, and the effort of understanding another person's perspective is fundamental. Bring AI in afterwards to find patterns, not during the first contact.
The final decision
When it is time to choose the problem statement or the concept, close the laptops. That decision has to be owned by the team, out of their debate, values and intuition. AI can lay out options; humans make the judgment call. This is the struggle that builds conviction.
The group synthesis
When a group is on the verge of a breakthrough, let them wrestle. The moment where one person's idea builds on another's and they arrive somewhere together is the peak of creative confidence. Reaching for AI there robs them of "we figured it out."
The personal reflection
Afterwards, ask people to reflect on their own: what were you proudest of, where did you get stuck, how did you get unstuck? That metacognitive work is what stops the session ending in "I don't know how we did it."
The format builds several of these off-ramps in already: every phase opens with human work before any tool, the problem statement is committed before boosts are offered, and the findings round asks the team what they think the record missed.