Figma commissioned its own randomized controlled trial, run with the research firm MeasuringU, to test what its prompt-to-design tool, Figma Make, actually saves people. Fifty product designers and fifty product managers did the same three tasks, half with the tool and half without — and the split in the results is stark: product managers saved 35% of their time, comfortably significant, while designers’ overall gain sits right on the edge of statistical significance and the paper’s own Discussion section contradicts its own Results section about whether that gain is real.
We go through what the trial actually measured — reproducing a design that was already decided, not the judgment calls designers are paid for — and where designers did see a real gain: the single hardest task, a comment-flyout prototype. We also cover the counterarguments from outside the trial, about what cheap, plausible AI output does to design review once nobody in the room feels sure enough to object.