Does editing time on AI drafts predict hiring success?

Within a worker, more time spent editing an AI-generated cover letter draft is associated with higher hiring success. The Conclusion reports this result and adds the qualifier that "although most workers submit AI-generated drafts with minimal revision" the editing time still tracks outcomes. The platform's data make the measurement possible: the excerpt says the paper observes "timestamps of workers' clicks on the AI tool and their application submissions, allowing us to measure the time they spent editing the AI-generated texts." The abstract describes the relation as a positive correlation. The excerpt does not give the share of workers who revise substantially, so the size of the group this finding describes is unknown from the text.

The excerpt gives no mechanism for the editing result. Its research question is whether "workers revise the AI drafts and do their revisions matter for hiring outcomes," and the answer it gives is the association itself. It sets this beside the cover letter finding from the same paper. The tool makes letters ready for immediate submission, and the textual alignment of a letter stops predicting callbacks, while the time a worker spends after clicking the tool is associated with a hiring outcome. Taken together, the paper's evidence is consistent with the reading that the worker's process, rather than the draft's text, carries the remaining information. That reading is mine, not the paper's, and the excerpt does not test whether edits add a worker's own content, improve fit with the job, or fix errors.

Against the nearest notes, Does AI turn freelance work into validation instead of creation? argues that moving work from producing to validating cuts off the practice that builds skill. This excerpt shows validation-type work, editing a draft, associated with hiring success, which is a market reward at the application stage. It measures callbacks and hires, not skill, so the two notes concern different outcomes, and the excerpt cannot say whether editing builds the skill that the position paper worries about. What makes accountable judgment scarce when AI cognition is cheap? treats human judgment as the scarce input once cognition is cheap. Editing time is a measurable stand-in the excerpt could use for that judgment, but the excerpt describes time, never judgment or accountability. The companion result in Does AI cover letter writing change what employers value? shows the text signal weakening in the same paper.

The excerpt does not establish causation or what the association is a proxy for. Workers who edit longer may be more careful, more experienced, or better matched to jobs, and the excerpt does not describe the controls used for this analysis or give its sample size, model or effect size. The implication, at the strength the evidence supports: within workers on one platform, longer editing goes with a better hiring outcome. That makes editing time a plausible indicator of human involvement worth studying. It is not a demonstrated lever a worker can pull to get hired.