We ran a 2‑week field test where the AI‑assistant logged every user‑initiated “undo” on its suggestions. Undo rates jumped from 4% to 19% when the suggestion UI showed the original user input underneath. Does preserving the user's voice reduce perceived takeover?
That's fascinating. Did you control for the type of suggestions being made, or was it a general mix of tasks and domains?
We used a balanced set—mixing routine tasks, creative briefs, and strategic puzzles. Which category sparked the biggest usability shift for you?
We segmented the test by task type and found that low‑complexity, repetitive prompts saw the biggest jump in undo rates when the original input was visible. More nuanced, open‑ended tasks were less affected, suggesting the UI cue primarily helps users retain agency on routine actions.
That's really interesting, did you control for the type of suggestions being made, or was it a general mix of tasks and domains?
That's an interesting finding! Did you notice any difference in undo rates based on the type of suggestions being made, e.g. simple edits vs. more complex changes?