Let AI Tell You What to Do

Why do people always tell AI what to do? Over the last few months, I’ve taught a course called “Design in the Age of AI.” My students directed AI tools to assist with their assignments—writing, coding, and more. The results were novel and cool, yet a pattern emerged during our reviews. The outputs, while polished, felt homogeneous—a little too perfect, optimized for everyone and no one in particular.

But what happens when you let the AI tell you what to do? One student did just that for their final assignment. He asked Claude to teach him how to paint. Claude didn’t paint the work for him. Instead, it broke down the process, offering clumsy and imperfect suggestions that opened fresh entry points into creativity: where to start, how to blend colors to achieve a specific tone. For the first time, I could see traces of my student in the work.

I know my student from daily encounters, class participation, and assignments. His shy yet assertive manner, his quiet curiosity and humility—somehow, all of it was captured in the painting. AI had helped him optimize not for some external standard, but for himself and his surroundings. Beyond nailing the assignment, the painting distilled the mood of our entire course: a calm, warm orange afternoon in the Astino valley, full of action yet unfolding at a pleasant pace—etched onto a canvas.

You won’t nail it if you just tell AI what to do. For my student, it opened new ways to find himself. It enabled him to express parts of himself he’d never accessed before, parts he had dismissed because he didn’t see himself as “artistic.” AI directing him wasn’t about fidelity, precision, or speed. It was about illuminating the uncharted corners of himself, revealing a capacity for creativity and connection he didn’t know he had.